<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Terence Faircloth on Medium]]></title>
        <description><![CDATA[Stories by Terence Faircloth on Medium]]></description>
        <link>https://medium.com/@terry.faircloth?source=rss-3a2f1afb56a4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*2Q73rRoPFvH8PBStGpZ11Q.jpeg</url>
            <title>Stories by Terence Faircloth on Medium</title>
            <link>https://medium.com/@terry.faircloth?source=rss-3a2f1afb56a4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 03:32:50 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@terry.faircloth/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Turn Any User Manual into a Personal Expert with Gemini Notebook]]></title>
            <link>https://medium.com/@terry.faircloth/turn-any-user-manual-into-a-personal-expert-with-gemini-notebook-3ca75c09b1de?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/3ca75c09b1de</guid>
            <category><![CDATA[notebooklm]]></category>
            <category><![CDATA[personal-user-manual]]></category>
            <category><![CDATA[productivity-hacks]]></category>
            <category><![CDATA[gemini-notebook]]></category>
            <category><![CDATA[user-manual]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Fri, 13 Nov 2026 02:37:16 GMT</pubDate>
            <atom:updated>2026-11-17T01:34:32.659Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*H8FQNMtKegzQRDFZJ10uiQ.png" /><figcaption>Image created by the author in Gemini Notebook</figcaption></figure><p>Most of us have had the same experience. A new device arrives, perhaps a robot vacuum, a smart thermostat, a camera, a pair of earbuds, or a hearing aid, and with it comes a user manual that is either a thin leaflet or a dense PDF of two hundred pages. Somewhere in that document is the answer to your question. Finding it means scrolling, guessing at the right search word, and often discovering that the feature you care about isn’t covered at all, because it lives in a companion app. There is a better way to work, and it uses a free tool from Google called Gemini Notebook.</p><p>Gemini Notebook is the new name for the research assistant many people knew as NotebookLM. The name has changed, but the central idea has not. Instead of answering from the whole internet, it answers from the sources you give it, and it shows numbered citations so you can check where each answer came from. That one feature makes it well suited to understanding a device. You decide what counts as the truth, and the notebook works only from that.</p><p><strong>Starting With the Manual</strong></p><p>The method begins with the official user manual. Nearly every manufacturer now publishes its manuals as PDF files on its support website, usually on a page you reach by searching for the model number. Download the manual and save it in a folder in Google Drive. I find it helpful to create one folder for each device and to name the file with the brand, the model number, and the date. Months later, there is no doubt which version I am looking at.</p><p>Next, open Gemini Notebook, create a new notebook named after the device, and add the manual from Drive as a source. Keeping your manuals in Drive does more than keep things tidy. According to Google’s help pages, <a href="https://support.google.com/notebooklm/answer/16215270">sources imported from Google Drive are kept in sync</a>, so if you later correct or replace a document in Drive, the notebook picks up the change.</p><p>There is a second, more important reason to work from Drive. Gemini Notebook reads the manual as text. It understands the words, but it doesn’t work with the diagrams, exploded drawings, button layouts, and wiring illustrations that make a manual truly useful. Because the original PDF sits in your Drive folder, you can go from the notebook’s source panel straight back to the full document and look at the picture that goes with the answer. I have tried this myself, and it works just as you would hope. The notebook tells me the reset button is described on page forty-two, and a moment later I am looking at the drawing that shows exactly where it is. This feature is not currently available in the mobile app, but you can access it on mobile devices by opening Notebook in a browser on your mobile device.</p><p><strong>Filling the App Gap</strong></p><p>This is where the method becomes genuinely useful. Many modern devices are really two products: the hardware in your hand and the app on your phone. The printed manual may tell you how to charge the device and press the power button. Everything interesting happens in the app: schedules, firmware updates, custom settings, sharing with family members. That app often has its own online help center, its own frequently asked questions, or a separate downloadable guide.</p><p>Gather that material too. If the app instructions exist as a PDF, save it in the same Drive folder and add it to the notebook. If they exist only as web pages, Gemini Notebook can take <a href="https://support.google.com/notebooklm/answer/16215270">web addresses directly as sources</a>, so you can paste in the help-center pages for the features you care about. Release notes for firmware and app updates are worth including as well, because they often describe features that appeared after the manual was printed.</p><p><strong>Adding YouTube Videos</strong></p><p>The third layer is video. For almost any popular device, someone on YouTube has recorded an unboxing, a setup walkthrough, or a “ten hidden features” video. These often explain things more plainly than the manufacturer does, and they cover real-world problems that official documents never mention.</p><p>Gemini Notebook accepts YouTube links as sources, but it helps to understand what it actually takes from them. Google explains that <a href="https://support.google.com/notebooklm/answer/16215270">only the text transcript of the video is imported</a>. The video must be public and have captions, either uploaded or automatically generated, and very recent uploads may not be available for about seventy-two hours. In other words, the notebook “hears” what the presenter says but does not “see” what the presenter shows.</p><p>That limitation matters less than it might seem, because the video itself is never far away. The notebook works from the transcript when it answers a question or writes notes, but the video stays available to you. In my own testing, clicking a video in the source panel opens a small window within that panel, and I can play the video right there to watch the step the notebook has just explained. If I want to see it at full size, a link on the face of the video takes me straight to the original on YouTube. The notebook finds the answer in the transcript, and the video shows the hands-on process. This feature is not currently available in the mobile app, but you can access it on mobile devices by opening Notebook in a browser on your mobile device.</p><p><strong>Other Sources Worth Adding</strong></p><p>Once you have the manual, the app guide, and a few videos, a few more additions can fill the remaining gaps. The manufacturer’s support page for your model usually has troubleshooting articles and a list of common questions. Enthusiast forums and community threads, added as web links, often contain fixes for problems the company has never acknowledged. An AI search engine such as Perplexity is a quick way to find the best of those threads before you add them.</p><p>Your own note is valuable too. Write a short document recording the model and serial number, the purchase date, the warranty terms, and any settings you have changed. Gemini Notebook also <a href="https://support.google.com/notebooklm/answer/16215270">accepts images</a>, so a photograph of the label on the back of the device, or of the quick-start card that came in the box, can become part of the collection.</p><p>There is a limit to how much you can add. Free users can include <a href="https://support.google.com/notebooklm/answer/16215270">up to fifty sources in a notebook</a>, and each source can be very large. For a single device, that is generous. The discipline lies less in adding more than in adding the right things.</p><p><strong>Putting the Notebook to Work</strong></p><p>With the sources in place, the notebook becomes a conversation partner that knows your device. You can ask in plain language, “How do I make the vacuum skip the bedroom on weekdays?” or “What do three blinking orange lights mean?” The answer draws on the manual, the app guide, and the videos together, with numbered citations that open the exact passage. When the manual and a video disagree, you can ask the notebook to point out the difference. And whenever an answer needs a picture or a demonstration, the PDF in Drive and the video are one click away.</p><p>The notebook can also produce new material from your sources. Google lists <a href="https://support.google.com/notebooklm/answer/16164461?co=GENIE.Platform=Desktop">study guides, briefings, audio overviews, and mind maps</a> among its outputs. For a device, that means you can ask for a one-page quick-start summary in your own words, a troubleshooting guide arranged by symptom, or a spoken overview to listen to while you set the device up. A mind map is a pleasant way to see at a glance how the features of a complicated gadget relate to one another. You can save these outputs and share them with a spouse, a grandchild, or anyone else who uses the device.</p><p><strong>Help Where the Problem Happens</strong></p><p>A notebook is most valuable when it is with you at the moment something goes wrong, and that moment isn’t always at your desk. Earbuds are a good example. Suppose you’re at the gym and only one earbud is playing, or a touch control keeps skipping tracks when you meant to change the volume. The four-page pamphlet that came in the box is sitting in a drawer at home, and it would be no help even if you had it. But your phone is already in your hand. Open the notebook in the Gemini Notebook mobile app and ask: “Only the right earbud is playing. Give me the simplest checks first, and don’t start with a factory reset.” You can ask for one step at a time, or for something short enough to read between exercises. Every answer comes from your instructions, for your model.</p><p>The lesson applies to any device you use away from home: hearing aids, a fitness watch, a travel camera. Building the notebook takes a little time at home, so the answer is ready when you’re somewhere else.</p><p><strong>Notebooks Inside the Gemini App</strong></p><p>Google has also brought notebooks into the main Gemini app. Its announcement explains that <a href="https://blog.google/innovation-and-ai/products/gemini-app/notebooks-gemini-notebooklm/">notebooks sync between the Gemini app and Gemini Notebook</a>, so a source added in one appears in the other. Inside the Gemini app, the assistant can also combine your chosen sources with web search. The rollout began with paid subscribers on the web, with free users and mobile access to follow, so availability may differ from one account to another.</p><p><strong>The Advantages</strong></p><p>The greatest advantage is that the answers are grounded. A general chatbot may confidently describe a button your model doesn’t have, because it is blending information about many similar products. A notebook built from your own sources is far less likely to do that, and when it does make a claim, the citation lets you check it in seconds.</p><p>The second advantage is unification. The manual, the app instructions, the videos, and your own notes no longer live in four places. One question searches them all, and the answer can combine a step from the manual with a tip from a video.</p><p>The third is plain language and context. Consider the humble “action button.” On a recent iPhone it is a small button on the side that can be set to open the camera or turn on the flashlight. On an Apple Watch Ultra it starts a workout or a stopwatch. On a camera, a remote control, or a kitchen appliance it may do something else entirely. Ask a general search engine what the action button does, and you’ll get a jumble of answers about devices you don’t own. Ask your notebook, and it answers for your device alone, because your sources are the only ones it consults. For readers who find dense technical writing tiring, being able to ask for a simpler explanation, or a fuller summary of a single topic, is a real kindness.</p><p>The fourth is portability. The notebook goes wherever your phone goes, which a paper manual never could.</p><p>Finally, the work is durable. The notebook stays available as long as you keep it. A year later, when the device misbehaves and the box has long since gone to recycling, your personal expert is still there.</p><p><strong>The Disadvantages</strong></p><p>The method has a few drawbacks. The notebook can still misread a passage, so check the original wording for anything involving safety or your warranty. Your sources go out of date as apps and firmware change, and automatic captions sometimes garble model numbers. Gathering good material takes some time for each device, and personal details such as serial numbers deserve care before you share a notebook. The whole system also depends on an internet connection and on Google keeping the service in its present form, which is one more reason to keep your original manuals safe in Drive.</p><p><strong>A Small Habit With a Large Payoff</strong></p><p>The real value of this approach is that it turns a frustrating chore into a routine. Each time a new device comes into the house, you spend a little time gathering its manual, its app instructions, and a few good videos. In return you get a patient, well-informed assistant that answers in plain language and shows its work, at home or anywhere else. The manual stops being a document you dread opening. It becomes the foundation of something much more useful: a personal guide built around the device you actually own and the questions you actually ask.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3ca75c09b1de" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your Books Have a Lot to Say with Gemini Notebook.]]></title>
            <link>https://medium.com/@terry.faircloth/your-books-have-a-lot-to-say-40bbce8dbceb?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/40bbce8dbceb</guid>
            <category><![CDATA[gemini-notebook]]></category>
            <category><![CDATA[notebooklm]]></category>
            <category><![CDATA[gen-ai-tools]]></category>
            <category><![CDATA[google-notebooklm]]></category>
            <category><![CDATA[ai-tools]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Thu, 05 Nov 2026 01:26:44 GMT</pubDate>
            <atom:updated>2026-11-05T17:35:21.079Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DATLLWe9Q-PSv1JK6FGXCw.png" /><figcaption>Image created by the author in Gemini Notebook</figcaption></figure><p>A good book does not finish its work when we finish reading it. We return for an explanation, a memorable passage, or an idea that suddenly matters in a new situation. But finding the right passage — and working out how it applies — can take time.</p><p>Google’s Gemini Notebook, formerly NotebookLM, offers another way into a book. You can question its contents, ask for a clearer explanation, and turn a chapter into study materials. Add your own notes and documents, and you can explore how the author’s ideas connect with the questions that brought you to the book.</p><p>The real promise is not faster summarizing. It is a closer connection between <strong>reading</strong>, questioning, and understanding: moving from a question to an explanation, then back to the passage that supports it. The notebook helps you explore; the judgment remains yours.</p><p><strong>Two Ways to Find Books</strong></p><p>Google offers two routes to books you can use in Gemini Notebook: a built-in catalog of free textbooks and a selection of eligible eBooks available through Google Play Books. They serve different purposes, and it helps to understand the distinction.</p><p><strong>The free textbook catalog</strong></p><p>To explore the textbook catalog, open a notebook, choose “Add sources,” and select “Textbooks.” The collection includes hundreds of titles from participating organizations, including OpenStax, Open Textbook Library, and S Chand Publishing. You do not need to purchase these books before adding them.</p><p>Subjects range from business, economics, and computer science to mathematics, history, writing, and psychology. The collection can be useful whether you are taking a course, returning to a neglected subject, or learning something entirely new.</p><p>Think of it as a selected educational bookshelf — not a complete collection of textbooks or of every title offered by the participating publishers.</p><p><strong>The Google Play Books selection.</strong></p><p>For reading beyond the textbook catalog, Google’s Expert Intelligence initiative lets you add eligible eBooks purchased through Google Play Books to a notebook. Google also provides a recommended selection to help you discover supported titles.</p><p>Start at <a href="https://notebook.google/expert-intelligence">Google’s Expert Intelligence page</a> and follow the links to explore books. Or begin inside Gemini Notebook: choose “Add sources,” select “Play Books,” and look for “Explore Google Play Books.”</p><p>Books in the textbook catalog can be added to Gemini Notebook free of charge. Google’s recommended selection, by contrast, highlights eligible eBooks available through Google Play Books. Being included in that selection does not make a book free: you may need to purchase it before you can add it to your notebook.</p><p>Before buying a book specifically for notebook use, check its individual Google Play Books page:</p><ul><li>Look for a “Tools” badge near the page count.</li><li>Select the badge and check whether Gemini Notebook is listed. The badge alone is not enough; it may describe other reading tools.</li><li>Check the particular edition you intend to buy.</li></ul><p>For a book you already own, open “About this book” and look for a “Use with Gemini Notebook” button. Eligibility depends on factors such as publisher participation, language, and the particular edition. Owning an eBook on another platform does not automatically make it available through this feature.</p><p>Begin with the free textbook catalog when you want to explore a subject without buying a book. Browse the Expert Intelligence recommendations when you want to discover other supported reading. If you already have a particular title in mind, check its eligibility before purchasing.</p><p><strong>Add a Book, Then Give It a Purpose</strong></p><p>For a free textbook, select a title from the built-in catalog and add it to your notebook. A “source” simply means material the notebook can draw on when answering your questions.</p><p>For a purchased book, sign in to Gemini Notebook with the same Google account you use for Google Play Books. Open or create a notebook, choose “Add sources,” select “Play Books,” and add an eligible title from your library.</p><p>Then give the conversation directions. A request to summarize an entire book may produce an orderly overview without helping you decide what to read or why. A more useful opening is:</p><p><em>“Give me a reading guide for an adult learning this subject independently. Explain the main ideas, identify the background knowledge I need, and suggest where to begin. Show me the passages supporting your explanation.”</em></p><p>Treat the response as a suggested route through the book, not a replacement for reading it.</p><p><strong>Ask Questions That Open Up the Subject</strong></p><p>Searching an eBook usually means looking for words or phrases. A conversation lets you begin with an idea — even when you do not yet know the author’s terminology.</p><p>With a history textbook, you might ask:</p><p><em>“Explain the main causes of the First World War in plain language. Separate the tensions that developed over many years from the events that immediately led to war. Show me the passages supporting your explanation.”</em></p><p>An introductory psychology textbook invites a different question:</p><p><em>“What does this book say about why we remember some things and forget others? Explain unfamiliar terms and give everyday examples.”</em></p><p>For economics:</p><p><em>“What is inflation, and why do prices rise? Explain the different causes discussed in this book, using simple examples rather than technical language.”</em></p><p>For biology:</p><p><em>“Explain how the immune system responds to an infection. Describe the main stages in plain language, explain unfamiliar terms, and show me the supporting passages.”</em></p><p>With a general-interest book, start with the author’s case:</p><p><em>“What is the author’s main argument? What evidence supports it, and what limitations does the author acknowledge?”</em></p><p>Google presents Expert Intelligence as a way to make books something like a personal brain trust. The distinction is worth keeping clear: you are asking an AI system to interpret published work, not receiving new advice from the authors themselves.</p><p>The notebook can link its answers to passages in the book. Those references make an explanation easier to check; they do not guarantee that the explanation is correct. An answer may miss an exception, misunderstand a qualification, or make a connection the author did not intend.</p><p>When an answer matters, follow the reference and read the surrounding text. Does the passage actually support what the notebook says?</p><p><strong>Bring Your Own Questions — and Your Notes</strong></p><p>The same book can serve very different readers. Someone preparing for an accounting examination needs a different explanation from someone trying to understand a small organization’s finances.</p><p>Tell the notebook what you want from your reading:</p><p><em>“I want a general understanding, not professional training. Explain unfamiliar terms, avoid assuming advanced mathematics, and distinguish the author’s statements from your own conclusions.”</em></p><p>You can also add a short document describing your background, interests, and questions. Include it among the sources the notebook uses.</p><p>Your notes can give the discussion a practical focus. Compare a management book with observations about a workplace problem. Read a history book alongside questions raised by another author. Bring a list of ideas you found confusing, persuasive, or incomplete.</p><p>But keep the origins of the information visible:</p><p><em>“Separate what comes from the book, what comes from my notes, and what you conclude by combining them.”</em></p><p>That distinction matters. A reference to your own notes shows where a statement came from; it is not independent evidence that the statement is true.</p><p><strong>Put Books in Conversation Without Making Them Agree</strong></p><p>A second book can do more than provide another summary. It can expose assumptions, fill gaps, or offer a different explanation.</p><p>When authors disagree, ask the notebook to preserve the disagreement rather than smooth it away:</p><p><em>“Compare these authors’ explanations. Identify where they genuinely disagree and where they use different terms or discuss different situations. Show the passages supporting each position.”</em></p><p>A tidy comparison can be helpful, but tidiness is not completeness. Return to the books when an important distinction seems to have disappeared.</p><p><strong>Turn Reading into Practice</strong></p><p>Gemini Notebook can turn source material into formats such as spoken overviews, flashcards, quizzes, video summaries, slide presentations, and mind maps.</p><p>Choose a format for what it helps you do. A spoken overview can introduce an unfamiliar chapter. Flashcards can help you review terminology. A mind map can help you examine connections. A quiz can reveal the difference between recognizing an explanation and being able to give one yourself.</p><p><strong>Use SWOT to examine an idea in practice.</strong></p><p>To move from understanding an idea to considering how you might use it, try a SWOT analysis: an examination of strengths, weaknesses, opportunities, and threats. Focus on a particular proposal or approach in the book rather than asking for a general assessment of the entire volume.</p><p>Suppose a management book recommends introducing a mentoring program. Add notes describing your organization, its resources, and what you hope to accomplish. Then ask the notebook to examine four questions:</p><ul><li>Strengths: What existing skills, resources, or relationships within the organization would help the program succeed?</li><li>Weaknesses: What internal limitations, such as insufficient staff time or experience, could undermine it?</li><li>Opportunities: What external partnerships, resources, or developments could support the program?</li><li>Threats: What outside pressures or changes could make success more difficult?</li></ul><p>These are questions to investigate, not findings to assume. You might combine them in a prompt:</p><p><em>“Using the book and my notes, develop a SWOT analysis of introducing a mentoring program in my organization. Separate internal strengths and weaknesses from external opportunities and threats. Explain each point and identify the supporting passages. Distinguish information in the sources from your own inferences and mark anything the sources do not establish as unknown.”</em></p><p>The purpose is not merely to fill four boxes. It is to examine what could make the proposal succeed, where it might fall short, and what you still need to investigate. Follow up with:</p><p><em>“How could we use our strengths to take advantage of the opportunities? Which weaknesses make us vulnerable to the threats? Suggest three next steps and explain what we should verify before acting.”</em></p><p>If you want to examine the author’s proposal without applying it to your own circumstances, make that clear:</p><p><em>“Analyze the proposal in this chapter. Identify its strengths and weaknesses, then examine the external conditions that could help or hinder its success. Support each point with passages from the book, distinguish the author’s statements from your interpretations, and leave a category incomplete if there is insufficient evidence.”</em></p><p>These activities change your role. Instead of collecting explanations, you are testing your understanding, weighing an author’s ideas, and considering how they apply to your circumstances. A polished summary is a starting point; the deeper value comes from what you can explain, question, and use.</p><p><strong>Prepare for Decisions; Do Not Outsource Them</strong></p><p>A book can help you prepare for a conversation with a specialist. It can explain vocabulary, clarify the principal issues, and suggest questions you would not otherwise know to ask.</p><p>Try:</p><p><em>“Using this book and my project notes, explain what I should understand before meeting a specialist. Suggest questions to ask and identify what these materials cannot establish.”</em></p><p>That is different from asking the notebook to approve a medical treatment, certify a contractor’s competence, or confirm that a financial decision is sound. Use it to become a better-prepared participant, not to supply an authority it does not possess.</p><p>The same caution applies to a SWOT analysis. A well-organized table does not establish that a proposal is workable. Its conclusions depend on the quality and completeness of the material you provide.</p><p>Books also age. On a changing subject, use a book to understand the underlying ideas, then verify current facts separately.</p><p><strong>Know What Does Not Come with the Book</strong></p><p>The book imports described here are text-only. Do not assume that the notebook has access to an illustration, diagram, table, or the original arrangement of a mathematical equation. When an explanation depends on visual material, consult the book itself.</p><p>This is especially important with worked examples. If an answer seems incomplete or puzzling, the missing information may be on the page rather than in the imported text.</p><p>Sharing has limits, too. Adding a purchased book does not make it freely distributable. Other people may need their own eligible copy to use it in a shared notebook, and publisher restrictions may affect what generated material you can download or share.</p><p>Before adding confidential notes or documents, check the current privacy terms for your account. Personal, organizational, and education accounts may have different protections, and submitting feedback may affect how associated material is handled.</p><p><strong>Begin with One Chapter</strong></p><p>For the first experiment, choose a chapter on a subject you already know something about. Your existing knowledge gives you a basis for judging the notebook’s explanations.</p><p>Ask it to explain one central idea. Read the passages it cites. Request another example. Then try explaining the idea yourself — or answer a few questions without looking back. If the chapter proposes something you might put into practice, use a SWOT analysis to explore its possibilities and limitations.</p><p>The aim is not to build the largest collection of summaries. It is to make your reading more useful: to find the passage you need, understand an idea that resisted you, or recognize a connection you had missed.</p><p>The best result is not a reason to put the book aside. It is a reason to return to it with better questions.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40bbce8dbceb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Mastering the Voice Features in Gemini Notebook]]></title>
            <link>https://medium.com/@terry.faircloth/mastering-the-voice-features-in-gemini-notebook-c93d16ee5a67?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/c93d16ee5a67</guid>
            <category><![CDATA[google-notebooklm]]></category>
            <category><![CDATA[notebooklm]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[gemini-notebook]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Tue, 06 Oct 2026 22:40:57 GMT</pubDate>
            <atom:updated>2026-10-06T22:40:57.221Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uaRvedPjnnJsu6SwPBpMBg.png" /><figcaption>Graphic created by the author in Gemini Notebook</figcaption></figure><p><strong>Moving beyond the voice memo</strong></p><p>Some of our best questions occur when we are nowhere near a keyboard. A teacher thinks of a better explanation while walking. A project manager leaves a meeting wondering whether a new commitment fits the agreed schedule. A student notices that a lecture and an assigned reading seem to offer different explanations.</p><p>A voice memo preserves the thought, but the inquiry often stops there. The recording remains on a phone, separate from the documents, readings, and notes needed to develop it. Capturing an idea is useful. Examining it is a different task.</p><p>Gemini Notebook, formerly Google’s NotebookLM, brings those activities closer together. Its audio capabilities allow users to capture spoken material, discuss notebook sources aloud, and listen to generated explanations. Google’s September 2026 announcement added mobile recording and real-time voice conversations to the product’s existing Audio Overviews.</p><p>The attraction is not simply another way to talk to an AI assistant. It is the opportunity to place a spoken question beside the evidence needed to investigate it. A recording can become the beginning of a working discussion rather than another file to find later.</p><p>That has value in professional work and education. A proposal can be compared with project documents. A student’s interpretation can be examined against a reading. A teacher’s explanation can be reviewed alongside lesson materials. In each case, recording the thought is the starting point, not the conclusion.</p><p><strong>Choosing the platform and feature</strong></p><p>Gemini Notebook offers three different uses of audio: recording material, discussing sources through Live Voice Chat, and listening to an Audio Overview. The device you use determines which of these activities is available.</p><p>The iPhone and Android apps support the built-in recorder and Live Voice Chat. The recorder adds material to the notebook; Live Voice Chat lets you ask spoken questions about material already there.</p><p>The desktop web version does not provide these native recording and Live Voice Chat controls. Its standard Chat interface uses typed questions. This distinction matters when following instructions: the Audio control described for mobile Chat is not a desktop control.</p><p>Audio Overviews are available on both mobile and desktop. You can generate and listen to a discussion based on selected sources and, where Interactive Mode is supported, join the hosts to ask a spoken question. The mobile app also allows downloads for offline listening.</p><p>The choice is therefore practical. Use the mobile app when you want to capture spoken material or conduct a live conversation with the notebook. Use either platform when you want a prepared audio discussion for orientation or review.</p><p><strong>Recording and saving a source</strong></p><p>On your iPhone or Android device, open the standalone Gemini Notebook app and select the notebook where the recording belongs. In the mobile interface described here, open Sources and select the recording control.</p><p>Three icons guide the process: a microphone to start, a pause symbol to suspend recording, and a check mark to finish and save.</p><p>Tap the microphone icon to begin capturing a lecture, meeting, or your own spoken notes. For a personal reflection, begin by naming the subject and stating the question. For example: “Reflection after the planning meeting. Does the proposed deadline allow enough time for testing?”</p><p>While recording, tap the pause icon — the two vertical lines — when you want to temporarily stop capturing audio. This is useful during a class break, a pause in a meeting, or a moment when you need to gather your thoughts. Resume when ready to continue within the same recording rather than create a separate file.</p><p>When finished, tap the check mark. Unlike pausing, this ends and saves the recording. Allow the app to finish processing the audio, then confirm that the resulting source appears in the notebook.</p><p>Review important names, dates, figures, quotations, and unfamiliar terms before relying on the transcription. When exact wording matters, retain the original recording and compare the relevant passage. Transcription should not be treated as a guarantee of word-for-word accuracy.</p><p>Give the source a title that identifies its role, such as “Lecture on historical interpretation,” “Project meeting,” or “My preliminary proposal.” A personal interpretation should remain distinguishable from a formal record or published reference.</p><p><strong>Starting Live Voice Chat</strong></p><p>To discuss the material, move from Sources to the Chat panel in the mobile app. Look for the Audio icon at the bottom right of the text-entry box.</p><p>On my iPhone, the icon resembles three lines and a star. Tapping it opens the live voice screen, and the microphone begins listening. You can immediately ask your question without typing or pressing a separate send button.</p><p>The animation in the center of the screen provides a visual indication of the session’s activity. Depending on the interface, it may appear as a waveform or a glowing shape. It is an indicator, not a button; you do not need to touch it to speak.</p><p>You guide the discussion naturally. If an answer is too long, unclear, or moving away from the question, interrupt. You might say, “Stop there and explain that distinction,” or “Which source supports that statement?”</p><p>The microphone button on the live voice screen lets you mute yourself without ending the session. This can help when background conversation or other noise is being mistaken for an interruption. Mute while listening, then turn the microphone back on when you are ready to ask another question.</p><p>When finished, tap Close, which may appear as an X or an end-session button. Transcripts of the exchange can be viewed in Chat and can be saved as a note. Live voice interaction is a mobile feature for users aged eighteen or older.</p><p>That transcript records your conversation with the assistant. It is different from a lecture or meeting transcript added to Sources. The recorder supplies material; Live Voice Chat helps you examine it.</p><p><strong>Creating an Audio Overview</strong></p><p>On either mobile or desktop, open Studio and select Audio Overview. Use the customization control to set the available options and explain what the discussion should emphasize. In the mobile interface, this control is shown with a pencil icon.</p><p>A focused request is usually more useful than a broad one. A professional might ask for a comparison of two proposals, concentrating on their assumptions and unresolved questions. A teacher could request an introduction for learners encountering a topic for the first time. A student might ask the hosts to explain where two authors disagree.</p><p>Once generated, the overview can be played. In the mobile app, it can also be downloaded for later listening.</p><p>An overview is prepared before you listen. Live Voice Chat, by contrast, develops through your questions and follow-ups. Neither approach is inherently better; they serve different purposes.</p><p><strong>Joining the Audio Overview</strong></p><p>Interactive Mode lets you move from listening to asking a question. It is available for supported English-language overviews on mobile and desktop and is described as a beta feature.</p><p>In the interface described here, generate an English-language overview and begin playback. Look for the waving-hand icon beside the play button and select it to enter Interactive Mode. When you hear something you want clarified, tap or click Join.</p><p>The hosts pause their discussion and acknowledge you. Wait for their invitation, then speak your question. A student might ask, “How does that explanation relate to the assigned reading?” A professional could ask, “Which evidence in the report supports that conclusion?”</p><p>When joining an Audio Overview in the desktop web version, you must use your <strong>voice</strong>. Even though you are on a computer, the Interactive Mode is designed specifically as an audio-in/audio-out experience. Here is how it works on desktop: (<strong>Microphone Required)</strong> when you click the <strong>Join</strong> button, your browser will activate your computer’s microphone. The AI podcast hosts will pause, and you must speak your question aloud to them.</p><p>There may be a short delay while the response is prepared. The hosts answer, then return to the overview.</p><p>The distinction from Live Voice Chat is straightforward. In live chat, you direct the exchange. In an interactive overview, the hosts lead a prepared discussion, and you join at particular points.</p><p><strong>Uses at work and in education</strong></p><p>The strongest applications keep the user’s own thinking visible. Voice should help identify uncertainty and examine evidence, not hide weak reasoning behind a fluent answer.</p><p><strong>Connecting lectures and readings</strong></p><p>A student could record a lecture, with permission, and add it to a notebook containing the assigned readings. After checking the transcription, the student could open Live Voice Chat and ask: “What does the lecture add to the reading? Is there a disagreement, or have I missed a distinction?”</p><p>An Audio Overview could then compare the materials more broadly. While listening, the student could join to ask about a particular point.</p><p>The purpose is not to accumulate summaries. It is to understand how the materials relate and improve the students’ own notes.</p><p><strong>Comparing meetings with written plans</strong></p><p>A professional could add a permitted meeting recording or a personal debrief alongside a project schedule and supporting documents.</p><p>A useful question would be: “Which commitments mentioned in this discussion differ from the written plan? Identify what needs clarification.”</p><p>Live Voice Chat can help examine individual discrepancies. An Audio Overview can provide a broader review before a follow-up meeting.</p><p>Neither should be treated as a ruling on what has been formally approved or what an agreement legally requires. A spoken commitment and an authorized change are not necessarily the same thing.</p><p><strong>Evaluating a proposal or interpretation</strong></p><p>Professionals often begin with an idea before assembling the evidence needed to support it. Students do much the same when developing an essay. Recording the idea can help make its assumptions explicit.</p><p>With relevant sources in the notebook, either user could ask: “Which parts of my argument have direct support? Which are assumptions? What evidence challenges my position?”</p><p>The next step belongs to the user: examine the passages, assess the objections, and decide what should change. The recorded reflection is the idea being tested, not independent confirmation of it.</p><p><strong>Preparing presentations and lessons</strong></p><p>A professional preparing a briefing can record an explanation and compare it with supporting documents. A teacher can do the same with a lesson plan and readings.</p><p>Useful questions include: “Where does this explanation move too quickly? Which terms need defining? Which claims need stronger support?”</p><p>Live Voice Chat provides a way to rehearse through questions and answers. An Audio Overview offers another perspective on how the material might be organized and explained.</p><p>The aim is not merely a smoother performance, but an explanation that remains clear when someone challenges it.</p><p><strong>Reviewing experience</strong></p><p>Short reflections can preserve lessons that are otherwise easy to lose. After a project milestone, a professional might record what worked and what caused delays. After a class, a teacher might note which explanation helped and which question revealed confusion.</p><p>Placed beside the original plan and relevant records, these reflections can support a focused review: “Where did the outcome differ from expectations? Which explanations are supported by the records, and which remain impressions?”</p><p>Over time, such notes can become a record of developing practice. Their usefulness depends on specificity and on avoiding unnecessary personal or confidential information.</p><p><strong>Keeping evidence and judgment together</strong></p><p>A workable process begins with a focused recording. Identify the subject, explain the thought, and state the question. Review the transcription, then add the material needed to investigate it.</p><p>Keep personal impressions distinguishable from formal records and published evidence. An instruction such as “Treat my recording as an interpretation to be tested” makes that distinction explicit.</p><p>Explore before requesting a finished document. Ask for comparisons, objections, alternative explanations, and unanswered questions. A report, study guide, or presentation draft may be useful later, but it should follow the examination of the material rather than replace it.</p><p>Return to the original sources before relying on an important conclusion. Check quotations and claims in context. A saved conversation does not establish that every answer was correct, and a lively Audio Overview is not necessarily a complete account. Generated overviews can contain inaccuracies.</p><p>Recording and uploading also require care. Permission to record a lecture, meeting, interview, or discussion should not be assumed. Check organizational policies, account protections, and sharing settings before adding confidential documents, student work, or identifiable speech. Different protections apply to personal accounts and Workspace or Workspace for Education accounts, but those protections do not establish that every upload is appropriate.</p><p>Before making an audio feature essential to a workplace process or classroom activity, check both the platform and the participating accounts. Mobile recording and Live Voice Chat should not be confused with the Audio Overview controls available in a desktop browser. Feature access and usage limits can also change.</p><p>A natural-sounding voice can make an answer feel more authoritative than it deserves. An explanation may be calm and persuasive while missing a contradiction or simplifying an important exception.</p><p>Gemini Notebook’s voice capabilities are useful when they make thinking easier to preserve and examine. Recording supplies material. Live Voice Chat helps question it. Audio Overviews offer another way to review it, with opportunities to join and ask for clarification. The value comes from connecting those activities to the evidence — and recognizing when a confident answer still needs checking.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c93d16ee5a67" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Put the Agents to Work for You in Gemini Notebook]]></title>
            <link>https://medium.com/@terry.faircloth/put-the-agents-to-work-for-you-in-gemini-notebook-da3a5ad15d49?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/da3a5ad15d49</guid>
            <category><![CDATA[notebooklm-tutorial]]></category>
            <category><![CDATA[gemini-notebook]]></category>
            <category><![CDATA[google-notebooklm]]></category>
            <category><![CDATA[agentic-ai]]></category>
            <category><![CDATA[notebooklm-strategies]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Tue, 15 Sep 2026 17:26:09 GMT</pubDate>
            <atom:updated>2026-09-15T17:26:09.822Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jyfrQOCQb2umG0Es4kxzKA.png" /></figure><p>Gemini Notebook, the product formerly known as NotebookLM, is no longer simply a place to upload documents and ask for summaries. Its newer software agents allow it to work through a connected series of actions: finding sources, comparing evidence, running calculations, organizing results and creating finished materials. Gemini Notebook can undertake much of the labor of research, but the user must still decide what question is worth asking, which evidence is trustworthy and whether the conclusions are justified.</p><p>The most useful way to understand agentic research is to imagine a very fast junior researcher working under supervision. A conventional chatbot often treats each prompt as a request for an immediate answer. An agentic system can instead interpret a larger assignment, divide it into stages, use appropriate tools and produce one or more working products along the way. The upgraded Gemini Notebook can use a secure cloud computer to write and run code and has a large collection of software skills for research and production tasks. That makes the system more capable, but not independent, infallible or self-validating.</p><h3>From answers to projects</h3><p>Suppose a user has collected municipal plans, engineering studies, meeting minutes and budget spreadsheets for a proposed public transportation project. A simple request such as “Summarize these files” will probably produce a competent overview. An agentic assignment can go much further: identify the major promises in the planning documents, compare them with later budgets, build a chronology, flag contradictions, calculate changes in projected cost and list questions that the available evidence cannot answer. The system is not merely composing a longer response. It is carrying out several related operations in service of a research objective.</p><p>This distinction explains why effective prompting is less about discovering a magic phrase than about defining a good assignment. Users generally do not need to command Gemini Notebook to “activate an agent” or specify a programming language. In Chat, the system can decide when web search, code execution or file creation is useful. A strong instruction identifies the evidence to examine, the sequence of work, the standards to apply, the desired result and, when appropriate, a point at which the system should stop for human review.</p><p>For example, a useful instruction might read:</p><p><em>“Using only the selected planning reports and budget files, first identify the project’s stated objectives and dates. Then compare projected and revised costs, explain how missing values were handled, and flag claims supported by only one source. Stop after producing an evidence table and a list of unresolved questions; do not draft the public briefing until I approve the evidence.”</em></p><p>This prompt invites analysis rather than a generic summary and builds a review gate into the process.</p><h3>The three working spaces</h3><p>Gemini Notebook is easiest to use when its interface is understood as three connected working spaces: Sources, Chat and Studio. Sources hold the evidence. Chat is the investigative and production workspace in which the user asks questions, directs actions, compares interpretations, checks citations and requests customized outputs. Studio provides dedicated tools for transforming the research into formats such as reports, data tables, slide decks, mind maps, flashcards, quizzes, and audio or video overviews.</p><p>The reliable sequence is therefore collect, interrogate, verify and produce. The final stage can take place either with Studio’s predefined formats or in Chat. Chat can generate a polished, customized article, report, table, file or other deliverable when it receives clear instructions about audience, purpose, form, length and tone. Studio is especially useful when one of its dedicated formats or controls suits the task. Whether the output is created in Chat or Studio, the important safeguard is to examine the evidence, expose disagreements and identify missing information before requesting the finished form. Polished language can conceal weak evidence just as easily as rough language can.</p><p>Source selection is an active part of the reasoning. Check only the documents that should govern a particular question and temporarily exclude unrelated material. Users can include or exclude individual sources with checkboxes, while citations can be opened to inspect the quoted passage in context. The notebook is therefore not merely a container; it is a controlled research environment whose boundaries can be adjusted from one question to the next.</p><p>Some capabilities described in this article may be absent or limited in the mobile app, so the desktop web version generally provides the fullest agentic research workflow.</p><h3>Two aspects of research</h3><p>Gemini Notebook agents can work inward, examining material already in a notebook, or outward, discovering new material. These are related activities, but they should not be confused.</p><p>When the necessary evidence is already present, the main work happens in Chat. The user selects the appropriate sources and gives a multi-stage instruction. Chat agents can answer questions about those sources, compare them, perform actions and create customized results, while its inline citations provide a route back to supporting passages. This source-grounded mode is especially useful when a carefully chosen collection should define the limits of the answer.</p><p>When additional evidence is needed, the Sources panel offers Fast Research and Deep Research agents. Fast Research searches the web or accessible Google Drive material for a focused set of candidate sources. It suits tasks such as locating a particular government report, finding recent documentation or checking whether an authoritative source exists. The user reviews the results and imports only the useful items.</p><p>Deep Research is intended for broader investigations. It can browse a large number of websites, reason through the findings and prepare a multi-page report. In the desktop procedure, the user enters a research question in the Sources panel, turns on Web and Deep Research, reviews the resulting report and candidate sources, and chooses what to import. Results that are not imported are discarded.</p><p>That review step is crucial. A source suggested by Deep Research is not automatically reliable merely because the system found it. Before importing, ask who created it, when it was produced, what evidence it contains, whether it is independent and whether a primary or more authoritative source is available. An industry association, a manufacturer, a university laboratory and a government regulator may discuss the same technology from very different positions. Those differences are part of the research, not an inconvenience to be smoothed away.</p><p>A well-framed Deep Research request should specify the subject, date range, preferred source types and competing perspectives. For instance:</p><p><em>“Research North American efforts between 2000 and 2025 to convert former industrial waterfronts into mixed-use public districts. Give priority to planning documents, environmental assessments, peer-reviewed urban-design studies and local economic-development reports. Compare claims about public access, housing affordability, environmental cleanup and private commercial development. Identify disagreements in reported costs or outcomes.”</em></p><p>Such a prompt gives the agent direction without asking it to decide the final interpretation in advance.</p><h3>The secure cloud computer</h3><p>One of Gemini Notebook’s most consequential additions is its secure cloud computer. For supported accounts and rollouts, the notebook agents can write and execute code for tasks such as cleaning data, calculating totals, finding patterns, creating charts and producing downloadable files. Some advanced functions remain dependent on subscription and rollout, so users should treat the controls visible in their own account as the practical guide to what is currently available.</p><p>Actual code execution can be more dependable than asking a language model to perform substantial arithmetic entirely in prose, but it does not guarantee a correct analysis. If a spreadsheet uses “N/A,” zero and a blank cell to mean different things — or perhaps to mean the same thing — the system can make a consequential mistake. Ambiguous headings, mixed date formats, merged cells, duplicate records and mismatched units can also lead to misleading results.</p><p>The proper protocol is to ask how the system interpreted the data. A careful request might say:</p><p><em>“Before calculating totals, list the columns you will use, identify missing and duplicate values, explain whether blanks are treated as zero or unknown, and show the formula for each derived field.”</em></p><p>Chat may expose details about the steps it is taking, but these should be regarded as an audit aid rather than proof of correctness. Important figures should still be checked against the original records, and a few sample calculations should be reproduced independently.</p><h3>Citations as verification tools</h3><p>Citations are among Gemini Notebook’s most valuable safeguards, but they work only when users click them. Chat responses can cite text and images from the selected sources; hovering reveals quoted material, and selecting a citation opens the relevant location so that it can be read in context. A citation therefore answers “Where did this statement come from?” It does not by itself answer “Is this source correct?” or even “Does the quoted passage fully support the wording of the claim?”</p><p>A good verification habit is to inspect the citation behind every important date, number, quotation and disputed assertion. Check whether the passage actually supports the claim, whether qualifications were omitted and whether the cited document is itself repeating another source. When several sources disagree, ask Gemini Notebook to display the competing accounts rather than choosing one silently. If the evidence is absent, the notebook should say so instead of filling the space with a plausible answer.</p><p>Source copies also require attention. Imported Google files may need to be manually resynchronized, and ordinary web imports may capture a page’s text without its embedded media or linked pages. Before relying on a frequently revised policy, price, schedule or shared document, confirm that the notebook contains the current version.</p><h3>Data tables that expose gaps</h3><p>Research evidence rarely arrives in a uniform format. One report may give annual costs, another monthly costs and a third only a narrative estimate. Gemini Notebook’s Data Tables can extract scattered facts into a common row-and-column structure, and the resulting table can be exported to Google Sheets. This is useful not only for presentation but also for diagnosis: blank cells and side-by-side disagreements are often easier to see in a table than in fluent prose.</p><p>Consider a household technology purchase. A notebook might contain product manuals, warranty terms, professional tests, manufacturer specifications and current retailer pages. The user could ask for a table comparing purchase price, required subscriptions, Windows and mobile compatibility, expected support period, warranty coverage and independently tested performance. The instruction should require the system to distinguish manufacturer claims from third-party test results, preserve disagreements and leave a cell blank when reliable information is unavailable.</p><p>That last requirement is important. AI systems tend toward the appearance of completeness, while good research often requires visible uncertainty. A blank cell labeled “not found” is more useful than an invented specification. Likewise, a tidy table should not erase the difference between a laboratory measurement, a manufacturer’s estimate and an anonymous customer review.</p><h3>Turning research into formats</h3><p>After the evidence has been tested, the user can request a customized final product directly in Chat or use Studio to recast the material through one of its dedicated formats. A detailed report may preserve nuance and documentation, while a slide deck may emphasize structure and decisions. A mind map can reveal relationships, and flashcards or quizzes can support learning. Audio Overviews present source material as a discussion between AI hosts and can be customized within Studio.</p><p>These are different representations of the same evidence, not independent confirmations. Audio Overviews reflect the sources and instructions supplied and are not necessarily comprehensive or objective. The same principle applies to every generated format. Weak, biased or outdated sources do not become reliable when converted into an engaging podcast, attractive infographic or polished presentation.</p><p>A sensible production workflow therefore separates substance from form without tying either stage to one part of the interface. First approve the evidence table, chronology or analytical notes. Then request the public-facing article, briefing or presentation in Chat or choose an appropriate Studio format. Finally, check that compression has not removed essential qualifications. An executive summary may need to be short, but brevity should not transform “one report estimated” into “research proves.”</p><h3>Everyday research projects</h3><p>Agentic methods are useful whenever a project contains many sources, repeated comparisons or several forms of output. In biographical or cultural-history research, a notebook can combine correspondence, newspapers, oral-history transcripts, census records and scholarly studies. It can trace professional relationships over time, compare contemporary documents with later memoirs and flag dates that appear in only one account. The resulting chronology is a map for further research, not proof that similarly named people have been correctly identified or that every reminiscence is accurate.</p><p>In consumer research, the notebook can cross-check advertising against manuals, warranty terms and independent tests. It can standardize specifications, identify compatibility problems and create several recommendations based on different priorities. Prices, availability, subscription terms and software-support promises should nevertheless be rechecked immediately before purchase because imported web material may become stale.</p><p>For civic associations, nonprofit boards or committees, Gemini Notebook can compare meeting minutes, policy documents, estimates, invoices and budgets. It can reconstruct when a decision was made, who was assigned responsibility and whether later records show completion. Before uploading confidential records, however, users should confirm their organization’s privacy, legal and contractual requirements rather than assuming that any convenient AI workspace is an approved repository.</p><p>In education and professional learning, a notebook can explain the same material at different levels, create study aids and connect claims to specific source passages. A learner might first request a plain-language explanation of a technical paper, then ask for a more advanced treatment of its methodology, followed by practice questions. A generated literature review or technical explanation should still be treated as a research aid: quotations, bibliographic details and interpretations require checking against the originals.</p><p>For a personal archive, the notebook can help identify recurring people, places and dates across letters, clippings, transcripts, photographs and notes. It may suggest tags, draft a chronology and flag uncertain authorship. Gemini Notebook should remain a research layer over a properly backed-up archive, not the sole home of irreplaceable files.</p><h3>A dependable protocol</h3><p>A reliable agentic project begins with a sharply defined question. Decide what the final result must help someone understand or decide, then assemble a source collection appropriate to that purpose. Select only the relevant sources for each stage, and use Fast or Deep Research when the existing collection lacks necessary evidence. Review provenance before importing anything.</p><p>Next, ask for an evidence product rather than a polished conclusion. A chronology, claim-and-source table, list of contradictions or data-quality report makes the system’s work easier to inspect. Tell Gemini Notebook to distinguish fact from interpretation, identify claims supported by only one source and state what it could not find. If calculations are involved, require an explanation of assumptions, missing values, units and formulas.</p><p>Then pause. Open the important citations, read the supporting passages and compare them with the notebook’s wording. Verify consequential numbers independently and investigate disagreements rather than allowing the system to average them into a false consensus. Refresh or re-import sources that may have changed.</p><p>Only after that review should the notebook create the final article, report, spreadsheet, presentation or Audio Overview. A custom finished product can be requested in Chat by specifying the audience, length, tone, structure and required qualifications; Studio can be used when its specialized output formats are more suitable. Read the finished work once for factual accuracy and again for distortion introduced by compression or presentation. Keep the original files in a separate, backed-up storage system and preserve enough notes to reconstruct how the final conclusions were reached.</p><h3>Productive skepticism</h3><p>Agentic research changes the scale of what an individual can attempt. A user can move from a question to a source collection, from scattered evidence to an organized analysis, and from analysis to several usable formats without manually performing every intermediate operation. Yet the increase in speed makes supervision more important, not less. A system that can produce ten pages, six charts and a slide deck in minutes can also spread one misunderstood column or weak source through all of them.</p><p>The best users will neither dismiss Gemini Notebook as an unreliable chatbot nor treat it as an autonomous expert. They will use it as a transparent, directed and revisable research partner. The human supplies purpose, judgment and accountability; the agent supplies speed, memory, organization and tool use. When those roles remain clear, agentic research becomes not a substitute for careful inquiry but a practical way to extend it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=da3a5ad15d49" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Practical Guide to Gemini Notebook Outputs]]></title>
            <link>https://medium.com/@terry.faircloth/the-practical-guide-to-gemini-notebook-outputs-c4c4380f9553?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/c4c4380f9553</guid>
            <category><![CDATA[gemini-notebook]]></category>
            <category><![CDATA[google-notebooklm]]></category>
            <category><![CDATA[notebooklm]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Mon, 14 Sep 2026 05:24:06 GMT</pubDate>
            <atom:updated>2026-09-14T06:14:11.939Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zMjy3KKeWQl8c8ZKUVp53Q.png" /><figcaption>Graphic created by the author in Notebook Gemini</figcaption></figure><p>Gemini Notebook, formerly NotebookLM, has two closely connected work areas: Chat and Studio. They are not separate worlds, nor is the difference simply that Chat is for research and Studio is for finished work. Both help me examine sources, organize findings, and create outputs that I may want to retain.</p><p>The most useful distinction is that Studio offers a set of predefined, customizable output formats, while Chat is the open-ended conversational workspace. In Studio, I begin by selecting a format — such as an Audio Overview, Report, Data Table, Mind Map, Slide Deck, Quiz, or Infographic — and then steer it toward a particular purpose. In Chat, I can begin with almost any instruction, revise it through follow-up requests, and decide what form the final result should take as I work.</p><p>That difference is not a difference in permanence. A Word document, spreadsheet, presentation, chart, or data file produced in Chat can be retained alongside Studio outputs in the notebook’s artifact list. Chat and Studio are therefore best understood as connected parts of a persistent research environment: one is more free form; the other begins with useful built-in formats.</p><p><strong>Source Grounding and Scope</strong></p><p>Both Chat and Studio draw on the sources added to the notebook: uploaded documents, Google files, web pages, pasted text, Notes, and media transcripts. Their usefulness depends on the quality, relevance, and boundaries of that source collection.</p><p>The source-selection controls are especially important. When every source is checked, Gemini Notebook can synthesize across the full notebook. That can be helpful for a broad orientation, but it may also bring together material that belongs in separate contexts. When I need a close comparison or a focused output, I select only the two, three, or four sources that bear directly on the question. This keeps the task narrow and makes it easier to see whether the answer is actually supported by the texts at hand.</p><p>Chat’s inline citations are particularly valuable here. They allow me to move from a statement back to the underlying passage, which is essential when I am checking a number, a date, a quotation, a subtle qualification, or a disputed interpretation. Studio outputs can be extremely useful for synthesis and presentation, but they should not exempt me from checking significant claims against the original sources.</p><p>A practical rule is simple: if the selected sources do not establish a fact, the output should acknowledge the gap rather than fill it with a plausible inference. I often make that expectation explicit in the request: “Use only the selected sources. Do not infer missing facts. Mark unsupported fields as ‘Not established in the sources.’”</p><p><strong>Chat: An Open-Ended Workspace</strong></p><p>Chat is the interactive center of the notebook. It is where I test claims, compare perspectives, locate direct quotations, identify contradictions, and work through an issue until I know what I need.</p><p>It is also a direct production environment. I can ask Chat to create a formatted Word document, PDF, Excel workbook, CSV file, JSON file, PowerPoint presentation, chart, diagram, or image. The benefit is not merely that it generates a file. It is that I can refine the file through conversation: change the title, shorten it, alter its intended audience, add columns, sort records differently, separate established facts from uncertain ones, or ask for a second version in another format.</p><p>For most requests, I specify three things: the file type, the contents, and the practical arrangement. If I need an Excel workbook, for example, I name the worksheet, define the columns, say how the rows should be sorted, and explain how missing data should be treated. If I need a Word briefing, I identify the audience, the desired headings, the approximate length, and the citation requirement.</p><p>Chat can produce several broad kinds of files:</p><ul><li>Documents, including Markdown, Word, PDF, and plain-text files.</li><li>Structured data, including CSV and JSON.</li><li>Spreadsheets, including Excel workbooks.</li><li>Presentations, including PowerPoint files.</li><li>Visual outputs, including PNG and SVG charts, diagrams, and images.</li></ul><p><strong>File-generation templates</strong></p><p>For a spreadsheet:</p><p>“Create an Excel spreadsheet titled <em>[Title]</em>. Include the columns <em>[Column 1]</em>, <em>[Column 2]</em>, <em>[Column 3]</em>, and <em>Source Citation</em>. Put one entity on each row, sort chronologically by <em>[Date Field]</em>, and leave a cell blank whenever the selected sources do not explicitly confirm the information.”</p><p>For a Word briefing:</p><p>“Create a Word document titled <em>[Title]</em> for <em>[Target Audience]</em>. Organize it under the headings Context, Key Findings, Methodological Discrepancies, and Unresolved Questions. Keep it to approximately 1,200 words and include source citations for factual assertions.”</p><p>For a JSON extraction:</p><p>“Generate a valid JSON file containing an array of objects, one for each <em>[Item]</em> identified in the selected sources. Each object must include id, title, creator, date, summary, and citation. Use null for a field that is not explicitly supported by the sources.”</p><p>For a CSV data matrix:</p><p>“Create a CSV file with the columns <em>[Entity]</em>, <em>[Attribute A]</em>, <em>[Attribute B]</em>, and <em>[Page or Source Reference]</em>. Use one row per entity. Output only the CSV data, with no explanatory preface.”</p><p>For a presentation:</p><p>“Create an eight-slide PowerPoint presentation titled <em>[Title]</em>. Include a title slide, an overview, four content slides, a conclusions slide, and a sources slide. Use concise bullets, readable headings, and a restrained visual design.”</p><p>For a chart:</p><p>“Create a PNG bar chart titled <em>[Title]</em> from the figures in the selected sources. Label both axes, include units, use readable category labels, and add a source note below the chart. Do not include values that cannot be established from the sources.”</p><p>These are production prompts, not research prompts. They assume that I have already selected appropriate source material and decided what I need to make. The purpose is to tell Chat exactly how to turn that material into a practical file.</p><p><strong>Studio: Predefined and Steerable Formats</strong></p><p>Studio is especially useful when one of its built-in formats already suits the job. Instead of designing an output entirely from scratch, I choose a format with an established structure and then direct it toward my audience, theme, and purpose.</p><p>The available formats include Audio Overviews, Video Overviews, Reports, Data Tables, Mind Maps, Slide Decks, Infographics, Flashcards, Quizzes, and Notes. These formats do not eliminate judgment; they give it a starting structure. A Data Table encourages me to think in fields and rows. A Slide Deck requires sequence and emphasis. A Mind Map brings relationships and hierarchy into view. An Audio Overview invites a broad, conversational understanding of a large body of material.</p><p>Studio formats can also be steered. Before generating an output, I can often set its focus, level of detail, tone, audience, language, and other parameters. The result is not simply a generic one-click summary. I can ask an Audio Overview to emphasize methodological disagreement, direct a Report toward operational risks rather than general background, or use a Data Table to expose missing information across several documents.</p><p>For example, I might instruct an Audio Overview:</p><p>“Focus the discussion on the methodological differences identified in Source 3. Maintain a critical, analytical tone for an audience of specialists, and do not spend time on general background.”</p><p>Or I might guide a report:</p><p>“Create a briefing document on the supply-chain vulnerabilities identified in the selected quarterly reports. Organize the findings by operational-risk level, omit general market background, and distinguish documented problems from projected risks.”</p><p><strong>What the Studio formats do</strong></p><p><strong>Audio Overviews</strong> are useful for orientation. The AI hosts discuss, connect, and explain the sources in a conversational form. I find them helpful when I want a broad understanding of a notebook before deciding which issues need closer examination. They are not a substitute for verifying precise wording, dates, or contested claims.</p><p><strong>Video Overviews</strong> provide narrated visual explanation, combining spoken commentary with generated visuals, diagrams, quotations, and other on-screen material. They can be useful when a process, place, timeline, or visual relationship matters as much as the prose explanation.</p><p><strong>Reports</strong> offer a defined written structure, such as a study guide, briefing document, FAQ, or custom report. They are useful when I know I need an organized synthesis but do not need to develop every aspect of its arrangement through a prolonged Chat exchange.</p><p><strong>Data Tables</strong> provide a structured way to extract information across sources. They work well for chronologies, filmographies, bibliographies, cast lists, issue trackers, evidence matrices, and comparisons. A well-designed table also makes absence visible: an empty or “not established” cell may tell me that the evidence is incomplete, rather than that the fact is known.</p><p><strong>Mind Maps</strong> make conceptual structure visible. They are useful early in a project, when I want to see themes, causes, categories, and relationships before I turn them into a linear argument or a piece of writing.</p><p><strong>Slide Decks and Infographics</strong> are useful when I need to communicate with an audience. A deck helps organize a progression of ideas; an infographic concentrates attention on what should be grasped at a glance.</p><p><strong>Flashcards and Quizzes</strong> support active recall. Flashcards are useful for names, dates, terms, and other discrete material. Quizzes can test broader comprehension by asking whether I understand relationships, arguments, and distinctions.</p><p>Studio is also a useful home for the notebook’s accumulated work. It contains its own built-in outputs, but it can also display files created through Chat. A Chat-created spreadsheet or Word document can therefore sit beside a Studio-generated report, mind map, or audio overview. The route by which I made an item does not determine whether I can keep and return to it.</p><p><strong>From Chat to Note to Studio</strong></p><p>One of the most productive ways to use Gemini Notebook is to treat Chat, Notes, and Studio as a continuous loop rather than as alternatives.</p><p>I may begin by selecting a small group of dense or difficult sources and using Chat to isolate the relevant claims, conflicting dates, recurring themes, or key quotations. Once I have a concise, cited synthesis that I trust, I can save it as a Note. That Note becomes a durable part of the notebook and can itself be used as grounding material for later work.</p><p>This can be especially helpful when the original sources are large, repetitive, or uneven in quality. Rather than ask Studio to work indiscriminately across hundreds of pages, I can first prepare a focused Note containing the evidence and distinctions that matter. I can then select that Note, alone or with the supporting documents, and use Studio to generate a targeted Audio Overview, Slide Deck, Mind Map, Report, or Data Table.</p><p>The sequence is not mandatory, but it is often effective:</p><ol><li>Select the relevant sources.</li><li>Use Chat to extract, compare, and clarify.</li><li>Save a strong, cited synthesis as a Note.</li><li>Use the Note and supporting sources to generate a Studio format.</li><li>Return to the original material when I need to audit the result.</li></ol><p>A Note should not be treated as a substitute for the underlying evidence. It is a useful working layer: a curated statement of what I have found, retaining citations and making subsequent outputs more focused. For significant claims, I still return to the original documents.</p><p><strong>Choosing Where to Start</strong></p><p>I start with Chat when my main need is flexibility. If I want to know exactly what selected sources say, compare conflicting figures, trace the language of a quotation, or refine a spreadsheet through several revisions, Chat is usually the better starting point. The same is true when I want a custom Word document, Excel workbook, or presentation whose structure will emerge through several practical adjustments.</p><p>I start with Studio when a predefined format will help me think or communicate. An Audio Overview is useful for a quick orientation to a broad notebook. A Mind Map can show conceptual structure before I know what argument to make. A Data Table is useful when the task is systematic extraction. A Report is useful when I need an organized synthesis. Flashcards and Quizzes are useful when the aim is review rather than writing.</p><p>Often the best path is mixed. I may begin with an Audio Overview to identify the main themes, move into Chat to verify specific claims and quotations, save the clarified synthesis as a Note, create an Excel workbook through Chat, and then use Studio to make a Slide Deck or Mind Map from the refined material. Each step has a different purpose, but all of them belong to the same research process.</p><p><strong>Verifying Outputs</strong></p><p>A polished layout does not guarantee accurate extraction. Audio Overviews, reports, diagrams, charts, slide decks, and summaries can make tentative inferences sound more settled than the sources justify.</p><p>Before I share, cite, or act on an output, I check the following:</p><ul><li><strong>Source selection.</strong> Did the result use the intended sources, or should the source checkboxes have been narrowed?</li><li><strong>Evidence and inference.</strong> Does the source state the conclusion directly, or has Gemini Notebook inferred a causal link, motive, chronology, or judgment?</li><li><strong>Missing data.</strong> Does an empty cell reflect an actual absence in the source, or did the system overlook an available detail?</li><li><strong>Quotations and attribution.</strong> Is the quotation exact, sufficiently contextualized, and attached to the correct person or document?</li><li><strong>Numbers and visualizations.</strong> Does a chart preserve the source figures, units, date ranges, and category definitions?</li><li><strong>Audience and purpose.</strong> Is the output appropriate for its use? An Audio Overview may be excellent for orientation but inadequate as a formal source record.</li></ul><p>When I need a formal reference table or a high-stakes summary, I make the constraints explicit: “Do not extrapolate. If a fact is not directly established in the selected sources, mark it ‘Not established.’ Include source citations for every factual assertion.” That instruction improves both Chat-created files and Studio outputs.</p><p><strong>Conclusion</strong></p><p>Gemini Notebook works best when Chat and Studio operate as a unified system. Chat is the open-ended, citation-backed workspace where I can investigate, clarify, revise, and create tailored files through conversation. Studio provides customizable, predefined formats that help turn source material into reports, tables, overviews, maps, presentations, visuals, and study tools.</p><p>The practical distinction is not one of permanence. Files created through Chat can remain available in the notebook alongside Studio-generated outputs. The real choice is whether I want to begin with an open-ended conversation or with a predefined format. By selecting sources carefully, saving strong syntheses as Notes, using Studio’s steering options, and auditing the results against the original material, I can turn Gemini Notebook from a source repository into a flexible and rigorous research workspace.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c4c4380f9553" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beyond the Photograph: Using Gemini Notebook to Read Art]]></title>
            <link>https://medium.com/@terry.faircloth/beyond-the-photograph-using-gemini-notebook-to-read-art-6437fe7d7d67?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/6437fe7d7d67</guid>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Thu, 10 Sep 2026 06:07:48 GMT</pubDate>
            <atom:updated>2026-09-10T23:28:00.670Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mXo9PIU69nd53OWg-BpwQQ.png" /></figure><p>I have started to use Gemini Notebook — formerly known as NotebookLM — not simply to summarize documents, but to work through visual evidence. That matters whenever a project begins with photographs: public murals, historic buildings, artworks, plaques, maps, archival scans, diagrams, or a collection of images gathered over several visits.</p><p>Unlike a conventional AI chat, where an image can feel like a temporary attachment, Gemini Notebook allows visual material to become part of a continuing research collection. I can place photographs beside articles, artist statements, field notes, city records, exhibition materials, and web clippings, then ask questions across the selected sources. The image becomes one item in a curated body of evidence rather than a single prompt that disappears into a chat history.</p><p>The key discipline is to distinguish between what an image establishes and what it merely suggests. A photograph may clearly show a painted jaguar, a gold-toned mask, a QR plaque, or a mural’s relationship to a street corner. It may suggest cultural influences, historical references, or artistic intention. But a photograph alone rarely proves an artist’s purpose, a work’s date, the identity of a figure, or the exact meaning of a symbol. Gemini Notebook is most useful when it helps preserve that distinction rather than turning a plausible interpretation into a fact.</p><p><strong>Use Case — A Walk Through Holland’s Murals</strong></p><p>Holland, Michigan, offers a good example of how this kind of visual research can work’ it has an annual festival, the <strong>Tulip City Walls Mural Festival</strong>. Its mural landscape turns blank walls and ordinary commercial buildings into vivid statements about cultural heritage, animals, local identity, and, in one instance, digital interactivity. Photographs of these works can be studied on their own, but they become more valuable when considered alongside city material, festival information, artist biographies, social-media posts, interviews, and notes made while walking through downtown.</p><p>For three murals, I uploaded to Gemini Notebook an image in a jpg format and a brief note with the artist name, Instagram handle, and street address.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-4fHhcosUR_xrfVcUnKKJQ.jpeg" /><figcaption>Photo by James Faircloth (@urbanmuralhunter), edited by the author</figcaption></figure><p>Kevin Burdick’s <em>Baile Folklorico</em> is a useful starting point. The mural depicts male and female dancers in traditional attire. The woman’s bright, multicolored skirt immediately commands attention, while the male dancer’s clothing includes an image associated with Our Lady of Guadalupe. An eagle, rattlesnake, desert cacti, and saturated blue and pink colors contribute to an elaborate visual setting.</p><p>Gemini Notebook gave the following analysis: <em>Baile Folklorico, The mural at 428 Columbia Avenue by Kevin Burdick highlights the “Baile Folklorico” tradition. It depicts a male and female dancer in traditional attire. The female dancer’s vibrant, multi-colored skirt is a focal point, while the male dancer’s clothing features an image of Our Lady of Guadalupe, signifying the deep-rooted connection between faith and folk culture. </em><strong><em>National Symbolism</em></strong><em>: The Baile Folklorico piece also incorporates the iconic imagery of a rattlesnake and an eagle, traditional symbols found on the Mexican flag, set against a backdrop of desert cacti.</em></p><p>Gemini Notebook can produce a careful inventory of those visible elements: figures, clothing, colors, animals, plants, and the way they are arranged. It can also help identify compositional features, such as the mural’s focal point, balance, scale, and contrast. Yet the next step requires care. It may be reasonable to read the work as bringing Mexican folk dance, national symbolism, and religious imagery into conversation, but that is an interpretation. Stronger claims about meaning should be connected to an artist’s statement, a curator’s text, or another credible source.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W2qs1mdJB6p3DJaRiScFhQ.jpeg" /><figcaption>Photo by James Faircloth (aurbanmuralhunter), edited by the author</figcaption></figure><p>TUKE’s <em>Box Turtle</em>, at 165 East 17th Street, raises a separate set of questions. The mural features a greatly enlarged turtle moving through tall grass and oversized red and yellow flowers. Its scale gives a familiar regional animal unusual authority and presence. A black plaque with a QR code identifies the work as an augmented-reality mural, adding a technological element to an otherwise painted public surface.</p><p>Gemini Notebook gave the following analysis: <em>The Box Turtle mural by TUKE focuses on a single, highly detailed turtle navigating through a field of oversized red and yellow flowers and tall grass. This piece emphasizes the beauty of regional fauna.</em></p><p>Here, Gemini Notebook can distinguish between observation and unanswered questions. A photograph can confirm that a QR plaque exists and that it labels the work as augmented reality. It cannot establish what happens when the code is scanned, whether the digital experience remains active, what device or app is required, or whether the interactive content has changed since installation. Those questions require testing the QR code and finding documentation from the artist, sponsor, or public-art organization.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZdWJjHcap97jkx8Hu7ZNoA.jpeg" /><figcaption>Photo by James Faircloth (@urbanmuralhunter), edited by the author.</figcaption></figure><p>Bruno Smoky’s mural at 408 Columbia Avenue, often identified as <em>Jaguar and Birds</em>, offers a third example. The imagery includes a jaguar, a macaw, a toucan, a prickly pear cactus, stylized botanical forms, and a central gold-toned mask. The work’s warm yellows, greens, and tropical motifs create a dense, layered scene. Its large animal forms and intense color demand attention even from a distance.</p><p>Gemini Notebook gave the following analysis: <em>Bruno Smoky’s work at 408 Columbia Avenue introduces tropical elements, including a jaguar, a macaw, and a toucan. The composition is rounded out by a potted prickly pear cactus and stylized botanical patterns, creating a dense, jungle-like atmosphere. </em><strong><em>Indigenous Roots</em></strong><em>, a prominent pre-Columbian or indigenous gold mask is centered, bridging the gap between ancient civilizations and contemporary visual culture.</em></p><p>Gemini Notebook can compare this mural with the others and identify patterns across a set of photographs: oversized animals, brilliant contrasting palettes, narrative layering, and the use of highly recognizable cultural or natural symbols. It can also point out meaningful differences. <em>Box Turtle</em> focuses on a regional creature in an enlarged natural setting, while <em>Jaguar and Birds</em> uses tropical wildlife and mask imagery to create a more densely symbolic visual world. <em>Baile Folklorico</em> puts human figures, costume, and cultural symbolism at the center.</p><p>This kind of comparison can generate better questions. Are the murals intended to form a coordinated collection, or are they independent works brought together by location? What organizations commissioned them? Are their references explained by the artists? How does the augmented-reality component alter the experience of seeing <em>Box Turtle</em>? A good visual-research system helps identify such questions; it does not pretend that every answer can be found in the pixels.</p><p><strong>What Gemini Notebook Can Do with Images</strong></p><p>Gemini Notebook works best when an image is treated as a source to be examined in relation to other sources. It can describe visible subjects and settings, identify broad compositional features, summarize colors and materials, note spatial relationships, and extract at least some visible text. With a collection of images, it can compare repeated motifs, changes over time, different views of the same object, or visual similarities and differences among artworks.</p><p>For a mural study, it might help identify recurring imagery such as animals, plants, masks, figures, flags, religious references, or graphic patterns. It can compare the treatment of color, scale, perspective, and surface. If photographs were taken at different dates, it could help flag visible signs of repainting, weathering, repairs, tagging, fading, or physical changes to the surrounding streetscape.</p><p>It is also useful for triangulation. An image can be placed beside an artist interview, local newspaper article, festival program, grant application, public-art map, or city planning document. Gemini Notebook can then help determine whether a written claim is visibly supported, whether it goes beyond the available image, or whether the sources disagree. For example, it can compare an artist’s description of a mural with what appears in photographs and identify details mentioned in the text that are not visible in the available documentation.</p><p>Visual material can also contribute to Gemini Notebook’s broader outputs. Depending on the features available to an account, images can inform written summaries, research briefs, study guides, mind maps, presentation material, audio or video overviews, and other syntheses. These can be useful ways to turn a scattered collection of photographs and documents into an initial catalogue, a chronology, a talk, or a list of questions for a return visit.</p><p>The most reliable prompts make the desired level of certainty explicit. I often ask for a response that separates directly visible observations from readable text, contextual interpretation, and unresolved questions. A prompt such as, “Describe only what is clearly visible, mark uncertain text rather than guessing, and identify claims that require a supporting written source,” encourages a much more dependable response.</p><p><strong>Image Types and Limitations</strong></p><p>Gemini Notebook can work with standalone image sources where they are supported by the current version of the service, as well as images embedded in documents such as PDFs, Google Docs, and Google Slides. In practical terms, ordinary photographs, phone images, screenshots, scans, maps, diagrams, charts, artworks, and pages from illustrated publications are all potentially useful research material. JPEG and PNG are generally the safest formats for individual images, while PDF is often the most practical choice for a sequence of scans, a contact sheet, or a visual dossier with captions and notes.</p><p>The tool can interpret a whole image and may be able to examine visible details within it, but image quality remains crucial. A close, well-lit, high-resolution photograph will yield far better results than a dark, distant, skewed, or heavily compressed one. When an inscription, signature, date, plaque, or small visual feature matters, it is wise to provide both a wide contextual image and a separate close crop.</p><p>Text recognition is helpful but fallible. Decorative lettering, handwriting, curved surfaces, glare, shadow, paint loss, reflections, perspective distortion, and low resolution can all produce inaccurate transcriptions. Important words, names, dates, and quotations should always be checked manually against the original image and, when possible, confirmed with an independent source.</p><p>Gemini Notebook should also not be treated as a specialist forensic, conservation, or authentication tool. It cannot reliably determine a painting’s exact pigment, prove an artist’s identity, authenticate an artwork, establish a precise date from visual style alone, diagnose structural damage, or identify a person or location with certainty. It may overlook small details, misread symbols, confuse visually similar objects, or produce an interpretation that sounds authoritative but goes beyond the evidence.</p><p>The best use of Gemini Notebook is therefore collaborative rather than automatic. Let it organize observations, compare images, extract tentative clues, connect pictures with documents, and surface contradictions. Then return to the original photograph, check the cited source, and verify important claims through artists, archives, official records, local organizations, or other trustworthy evidence.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6437fe7d7d67" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Talking to Your Bookshelf: How Expert Intelligence Brings Deep Knowledge to Gemini Notebook]]></title>
            <link>https://medium.com/@terry.faircloth/talking-to-your-bookshelf-how-expert-intelligence-brings-deep-knowledge-to-gemini-notebook-9930f29a353e?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/9930f29a353e</guid>
            <category><![CDATA[expert-intelligence]]></category>
            <category><![CDATA[notebooklm]]></category>
            <category><![CDATA[gemini-notebook]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Wed, 02 Sep 2026 05:51:37 GMT</pubDate>
            <atom:updated>2026-09-02T05:55:38.048Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*k5ZfYIAOJvizJey0Xpxf_g.png" /><figcaption>Image created by the author in Midjourney 8.2</figcaption></figure><p>For generations, reading a book has been a strictly one-way conversation. Readers highlight passages, dog-ear pages, and try their best to recall key frameworks weeks or months later. When an author’s insight is urgently needed to solve a specific problem, the exact quote or model is often buried somewhere across hundreds of pages. Google is reshaping this experience through <strong>Expert Intelligence</strong> in <strong>Gemini Notebook</strong>, the evolved rebranding of NotebookLM. By integrating professionally published literature directly with personal notes and files, Gemini Notebook transforms static reading into an interactive, grounded research partnership.</p><p><strong>Acquiring Books and Navigating the Catalog</strong></p><p>To bring a published volume into a research workspace, users purchase eligible digital editions directly through the Google Play Store or the Google Play Books web interface. While Google has launched the feature with more than 100,000 eligible titles from major publishing houses — such as Penguin Random House, Macmillan, Bloomsbury, O’Reilly Media, De Gruyter Brill, and Johns Hopkins University Press — there is currently no master storefront index or dedicated browse category listing all supported books in one place. As a result, readers must verify eligibility title by title by navigating to a book’s detail page and clicking the interactive <strong>Tools</strong> badge beneath the title. If the badge displays “Gemini Notebook,” the title supports AI integration. Once purchased, the book immediately binds to the user’s Google account and can be imported directly into any notebook without file conversions or external ePub uploads.</p><p><strong>Moving from Reading to Querying</strong></p><p>At its core, Expert Intelligence enables users to import these purchased e-books directly into a Gemini Notebook workspace as authoritative sources of truth. Instead of relying on general open-web training data, which can produce AI hallucinations or invented facts, the system uses retrieval-augmented generation (RAG) to scan the specific text of imported materials. Crucially, every single response, claim, and summary includes inline citations pointing back to the exact passage, chapter, or page. These direct references serve as an immediate verification check, ensuring the AI remains a strictly grounded assistant while the user stays in full control of the facts.</p><p><strong>Merging Published Books with Personal Knowledge</strong></p><p>Where traditional research tools isolate commercial books from private working notes, Gemini Notebook enables hybrid knowledge synthesis by allowing up to three hundred diverse sources in a single workspace. A corporate manager can pair internal strategy documents with Kim Scott’s <em>Radical Candor</em> to simulate difficult feedback conversations, while a software engineer can load proprietary code repositories alongside authoritative O’Reilly technical books to evaluate architectural trade-offs. Similarly, readers can compare personal health journals with Dr. Mary Claire Haver’s <em>The New Menopause</em> to prepare doctor-visit question lists, or match pantry inventories with Michael Pollan’s <em>Food Rules</em> to design weekly meal plans. In higher education, students can place lecture audio recordings and personal notes next to university press textbooks to generate tailored study materials and resolve contradictions between lectures and assigned readings.</p><p><strong>Multimodal Study and Visual Formats</strong></p><p>Expert Intelligence extends beyond simple text summaries by transforming ingested literature into interactive study aids. The system can generate conversational, podcast-style Audio Overviews featuring two hosts who unpack complex topics across single titles or synthesized collections of notes. For visual thinkers, the platform automatically creates concept mind maps and structural overviews that clarify complex hierarchies and workflows. To reinforce retention, it also constructs customized digital flashcards, and multiple-choice quizzes complete with explanatory answer keys.</p><p><strong>Licensing, Rights, and Shared Notebooks</strong></p><p>To protect intellectual property and support authors, Expert Intelligence operates under strict copyright and digital rights management controls. When sharing a collaborative workspace, collaborators can read the host’s personal notes and see the book listed as a source, but they cannot directly query or generate outputs from the published text. Instead, the interface prompts collaborators to purchase their own licensed copy via Google Play Books, ensuring that publishers and authors are compensated while keeping multi-user research legitimate.</p><p><strong>Privacy and Future Expansions</strong></p><p>Personal uploads, user queries, and generated outputs in Gemini Notebook are not used to train Google’s base AI models in business, educational, or standard enterprise tiers. For enterprise clients, all indexing and machine learning processing remain secured within the customer’s Google Cloud project boundary. Moving forward, Google plans to expand Expert Intelligence into the primary Gemini application and AI Mode in Search, while growing the underlying catalog to include third-party subscriptions, corporate business research reports, and specialized academic textbooks. By turning static volumes into interactive, cited knowledge engines, Expert Intelligence offers a reliable way to weave expert thinking directly into daily research and practical workflows.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9930f29a353e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beyond the Prompt: Restoring Evidence and Judgment in Business Education]]></title>
            <link>https://medium.com/@terry.faircloth/beyond-the-prompt-restoring-evidence-and-judgment-in-business-education-a891354fca5b?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/a891354fca5b</guid>
            <category><![CDATA[gemini-notebook]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[higher-education]]></category>
            <category><![CDATA[education-technology]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Mon, 31 Aug 2026 04:01:01 GMT</pubDate>
            <atom:updated>2026-08-31T04:15:45.798Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oC4wcOMPhmVHjLIbGbYnrg.png" /><figcaption>Image created by the author in Midjourney 8.2</figcaption></figure><p>Graduate business education now operates within a new instructional tension. Many students arrive with fluency in AI interfaces: they know how to move quickly, iterate prompts, and produce polished responses at impressive speed. Faculty, however, hold a different and more consequential form of expertise: judgment about which sources matter, which frameworks remain durable, what evidence can support a claim, and where apparent consensus masks uncertainty.</p><p>This distinction has become more important as open-ended generative AI has entered graduate classrooms. General-purpose systems can produce plausible recommendations, concise case summaries, and authoritative-sounding arguments in seconds. Yet fluency is different from rigor. A response may sound strategic, quantitative, or executive-ready while resting on incomplete, outdated, weakly sourced, or invented evidence.</p><p>Gemini Notebook provides a way to address that problem by making the quality of a notebook’s source base central to the AI interaction. Its newer <strong>Expert Intelligence</strong> capability extends this model by bringing selected expert and publisher-backed materials into Gemini Notebook’s source-grounded environment. Rather than treating AI merely as a prompt-responsive text generator, the platform can be used to create an instructional workspace organized around faculty-selected course materials and, where available and appropriately licensed, authoritative expert texts.</p><p>For graduate business schools, the significance is clear. The question is no longer simply whether students will use AI. They will. The more important question is whether AI use will be organized around speed and surface plausibility or around curation, traceable evidence, and disciplined interpretation.</p><p><strong>From prompting to curation</strong></p><p>The educational value of Gemini Notebook is not that it makes faculty into expert prompt engineers. Its value is that it allows faculty to define a bounded intellectual environment within which AI-assisted inquiry occurs.</p><p>A traditional notebook may include instructor-uploaded materials: syllabi, lecture slides, case packets, journal articles, working papers, data documentation, websites, recordings, transcripts, and student-facing guidance. Expert Intelligence expands the idea of the notebook by making it possible to incorporate qualified expert materials and professionally published sources into that environment where Google, publishers, licensing arrangements, and institutional access permits.</p><p>This is an important shift in emphasis. The central question becomes less, “What is the cleverest way to ask the model a question?” and more, “What body of knowledge should govern the model’s answer?”</p><p>For a business-school professor, that body of knowledge may include a combination of canonical management texts, current scholarly literature, institutional reports, teaching cases, lecture materials, and specialized expert sources. A strategy professor might combine classic work on competitive advantage with contemporary writing on platform economics, geopolitical risk, industrial policy, supply-chain resilience, and digital ecosystems. A finance professor might combine a core text with journal articles, case materials, professional standards, and curated expert commentary on valuation or capital allocation.</p><p>The resulting notebook does not turn AI output into unquestionable truth. It does, however, make the source base more visible, more deliberate, and more reviewable. That is a substantial improvement over an unrestricted workflow in which students receive polished answers whose evidentiary origins may be unclear.</p><p><strong>What Expert Intelligence adds</strong></p><p>Expert Intelligence should be presented not as a replacement for faculty curation, but as a potential enhancement to it. It can extend a course notebook beyond materials manually assembled by an instructor by allowing authoritative expert and publisher-backed content to become part of the notebook’s grounded research environment.</p><p>This is particularly valuable in business education because the field combines multiple forms of knowledge. Students must learn from peer-reviewed research, influential books, managerial cases, industry evidence, economic conditions, and professional practice. No single source type is sufficient.</p><p>An instructor using Expert Intelligence might build a notebook around three layers of material:</p><ul><li><strong>Course-specific sources</strong>, including the syllabus, slides, lectures, cases, assignment briefs, grading rubrics, and instructor notes.</li><li><strong>Scholarly and professional evidence</strong>, including peer-reviewed research, working papers, corporate filings, institutional reports, and high-quality industry analysis.</li><li><strong>Expert Intelligence materials</strong>, including available expert texts and publisher-backed works that deepen the course’s treatment of foundational concepts, historical context, or specialized practice.</li></ul><p>The advantage of this layered model is not simply greater volume. It is the possibility of putting classic concepts, current evidence, and expert explanation into conversation with one another. Students can ask not only what a case recommends, but how that recommendation aligns with established theory, where it departs from prior research, and which assumptions require validation in the present business environment.</p><p><strong>A curated course environment</strong></p><p>A well-designed graduate course notebook should be treated as an extension of the course architecture, not as a miscellaneous repository of files. Faculty should establish one dedicated notebook for each course offering, titled clearly by course, section, and term. It should contain a coherent and limited set of materials that students can reasonably understand as the evidence base for their work.</p><p>For example, a notebook for <strong>MGMT 801: Advanced Strategic Analysis — Fall 2026</strong> might include the syllabus, session slides, assigned cases, journal articles, lecture recordings, selected market reports, and licensed expert texts available through Expert Intelligence. The faculty member may also include a short course memorandum explaining how students should use the notebook: as a source-grounded study and analytical tool, not as a substitute for reading, independent judgment, or external research where an assignment requires it.</p><p>Source naming is more important than it may initially appear. Clear labels allow students to understand citations and return to the underlying material quickly. A source called “Porter Competitive Advantage Chapter 2” is more useful if labeled, for example, <strong>[Week 3] Porter, Competitive Advantage — Chapter 2</strong>. A case should identify the organization, issue, and teaching week. A lecture transcript should specify the date and topic. Good metadata is a modest administrative investment that can significantly improve citation literacy.</p><p>The notebook should also include explicit instructional guidance. Faculty can tell the system and students that its role is to support analysis, expose ambiguity, generate questions, and identify relevant passages — not simply deliver finished answers. A course might direct students to ask for competing interpretations, assumptions behind recommendations, counterarguments, and evidence gaps.</p><p>A useful default instruction could read:</p><p>Treat every business recommendation as provisional. Identify the specific sources supporting it, distinguish evidence from inference, name important assumptions, and identify information that would be needed before an executive decision could be justified.</p><p>That instruction captures an essential graduate-business-school norm. Strong analysis is not merely an answer. It is an answer with an evidentiary basis, stated assumptions, and a clear account of uncertainty.</p><p><strong>Preparing richer class discussion</strong></p><p>One of the most promising faculty uses of Gemini Notebook and Expert Intelligence is preparation for classroom discussion. Instructors can use the notebook to anticipate the conventional recommendations students may generate from a case and then design questions that reveal the limits of those recommendations.</p><p>Consider a case involving a consumer-products company deciding whether to expand internationally. An instructor might ask the notebook to identify the five most likely recommendations that an open-ended generative AI system — or a hurried student — would propose. Such recommendations might include entering the largest market first, reducing prices to accelerate adoption, forming a local partnership, increasing digital marketing, or centralizing supply-chain management.</p><p>The instructor can then use the course corpus, including expert texts available through Expert Intelligence, to formulate better questions:</p><ul><li>What assumptions about customer behavior, institutional conditions, and local competition underlie the recommendation?</li><li>Which frameworks in the assigned strategy literature challenge those assumptions?</li><li>What historical examples suggest that a standardization strategy may fail?</li><li>Which facts in the case contradict a rapid-entry recommendation?</li><li>What external market evidence would be required before approving the investment?</li></ul><p>The point is not to embarrass students for using AI. It is to teach them that a plausible first answer is often only the beginning of analysis. Expert Intelligence can deepen this preparation when it enables the notebook to draw upon recognized texts that supply context, conceptual history, or professional interpretation beyond the individual case packet.</p><p><strong>Improving case analysis</strong></p><p>The most constructive classroom use of Gemini Notebook is as a <strong>closed or bounded analytical environment</strong>. Within such an environment, students can ask questions about the assigned corpus and receive answers linked to the materials they are expected to study.</p><p>In a finance course, students might use the notebook to compare a case company’s capital-allocation decisions with principles drawn from assigned readings and expert materials. Instead of asking, “What should the company do?” they might be guided to ask:</p><p>Using the case exhibits, assigned valuation readings, and available expert sources, identify the three strongest arguments for and against the proposed acquisition. Distinguish direct evidence from assumptions.</p><p>In a leadership course, students might ask:</p><p>Compare the CEO’s stated rationale with the leadership frameworks in the assigned materials and Expert Intelligence sources. What organizational risks does the CEO appear to underestimate?</p><p>In an operations or supply-chain course, a student team might ask:</p><p>Based only on the case, course readings, and available expert sources, identify the trade-offs among cost, resilience, supplier concentration, and lead time. Which conclusion is best supported, and what evidence remains missing?</p><p>These prompts turn the notebook away from generic recommendation generation and toward disciplined reasoning. The student must connect evidence, frameworks, and judgment rather than merely request a finished managerial answer.</p><p><strong>Faculty archives and expert texts</strong></p><p>Long-serving faculty frequently possess rich collections that have become difficult to use in their original form: lecture notes, historical case annotations, research memoranda, guest-speaker transcripts, unpublished teaching materials, and decades of accumulated professional documents. Gemini Notebook can help transform these static archives into interactive academic resources when their use is appropriate and permissioned.</p><p>A professor of organizational behavior, for example, might combine archival lectures on organizational culture with current articles and Expert Intelligence materials from respected management thinkers. Students could then compare the professor’s earlier teaching on a subject with newer research or contemporary professional perspectives. This can reveal how durable concepts have evolved without treating either older or newer material as automatically superior.</p><p>The same process can strengthen executive education. A faculty member preparing a program for senior managers may assemble a source base that includes current company documents, industry reports, academic research, and expert texts relevant to the program’s subject. The notebook can then be used before and between sessions to help participants revisit concepts, identify tensions among sources, and prepare evidence-based questions for discussion.</p><p>The key principle is that Expert Intelligence should enrich a faculty member’s intellectual design. It should not become an excuse to abandon that design to a vendor-curated collection of texts.</p><p><strong>Assessment and academic integrity</strong></p><p>The arrival of AI in graduate education requires a shift in how faculty think about academic integrity. The goal should not be to determine, through unreliable detection methods, whether a student used an AI tool. The more meaningful question is whether the student can demonstrate responsible, transparent, and evidence-based use of that tool.</p><p>Gemini Notebook can support this shift because it makes source-grounded work more visible. Faculty can design assignments that require students to identify the sources underlying a claim, compare competing interpretations, explain why a cited source is relevant, and acknowledge where notebook-based analysis remains incomplete.</p><p>A strategy memorandum, for instance, might require students to include a short “evidence and assumptions” section. Students would identify the course materials or Expert Intelligence sources that support their recommendation, distinguish those sources from their own managerial inferences, and specify what outside evidence would be required to validate the recommendation before implementation.</p><p>Similarly, a capstone project could require an AI-use statement explaining whether Gemini Notebook was used for source review, brainstorming, comparison of frameworks, drafting support, or citation checking. The student would remain responsible for the argument, the accuracy of the citations, and the final written work.</p><p>This approach treats AI literacy as a form of professional judgment. In business practice, executives and analysts routinely use tools, consultants, data platforms, and research services. The ethical issue is not that assistance exists. It is whether the professional understands the provenance, limitations, assumptions, and decision consequences of the information being used.</p><p><strong>Governance and confidentiality</strong></p><p>No serious deployment of Gemini Notebook in a graduate business school should proceed without governance rules. The platform may provide meaningful institutional data protection, particularly when used through Google Workspace for Education, but privacy protections do not by themselves determine whether a document is appropriate for upload. Faculty and students must still comply with university policy, contracts, copyright restrictions, research protocols, and professional obligations.</p><p>Schools should clearly prohibit or restrict the upload of confidential consulting materials, client information, nonpublic corporate data, personally identifiable student records, restricted research materials, and case documents whose licenses do not permit the intended use. Students working in internships or employment settings should be reminded that an employer’s information is not automatically appropriate for use in a course notebook, even when the student has legitimate access to it.</p><p>The same care applies to Expert Intelligence materials. The availability of expert or publisher-backed content within a platform does not eliminate the need to understand licensing terms, institutional access requirements, permitted sharing, or limits on reuse. A course director or librarian should be involved when a program intends to make licensed materials available at scale.</p><p>This governance framework should be presented as an educational strength, not an administrative obstacle. Graduate business education should prepare students for environments in which data stewardship, intellectual-property judgment, and confidentiality are integral to professional competence.</p><p><strong>Limits and appropriate expectations</strong></p><p>Gemini Notebook and Expert Intelligence are best suited to source-intensive tasks: case preparation, literature review, comparison of frameworks, lecture review, discussion preparation, and the analysis of a defined body of evidence. They are not complete substitutes for current market databases, independently verified external research, financial modeling, legal review, or direct managerial investigation.</p><p>A notebook’s output is only as reliable as its source base, the relevance of the question, and the user’s capacity to inspect the answer critically. A citation demonstrates that a response has been linked to a source; it does not automatically prove that the source is current, complete, correctly interpreted, or sufficient for a high-stakes decision.</p><p>Faculty should therefore teach students to use the notebook in stages. First, identify the relevant evidence. Second, test competing interpretations. Third, state the assumptions connecting evidence to recommendation. Fourth, identify what remains unknown. Finally, determine which questions require research outside the notebook’s source boundaries.</p><p>That workflow has enduring value whether or not a student uses AI. Gemini Notebook simply makes it easier to bring the workflow into daily course practice.</p><p><strong>Reclaiming rigor</strong></p><p>The deeper promise of Gemini Notebook Expert Intelligence is not that it allows graduate business schools to make AI safer through technological containment alone. Its promise is pedagogical. It gives faculty an opportunity to make curation, source quality, provenance, and verification visible again at the point where students are forming arguments.</p><p>The most important AI skill for a graduate business student is not prompt cleverness. It is the ability to distinguish a persuasive answer from a warranted one; to recognize the difference between a citation and adequate evidence; and to know when an apparently confident recommendation rests on assumptions that have not yet been tested.</p><p>By combining faculty-curated course materials with appropriately available expert and publisher-backed sources, Gemini Notebook Expert Intelligence can help create a more disciplined form of AI-assisted learning. It can make the graduate classroom a place where artificial intelligence supports — not substitutes for — the habits of inquiry that business education should defend: careful reading, evidentiary judgment, intellectual humility, and accountable decision-making.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a891354fca5b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Turn Words into Action]]></title>
            <link>https://medium.com/@terry.faircloth/turn-words-into-action-0dbb7a13a143?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/0dbb7a13a143</guid>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Sun, 02 Aug 2026 01:13:20 GMT</pubDate>
            <atom:updated>2026-08-02T01:13:20.297Z</atom:updated>
            <content:encoded><![CDATA[<h4>From Research to Execution: Notebook Meets Perplexity</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AsyDi3z6DqGD1lXms_aeIQ.png" /><figcaption>Image created by the author in Midjourney 8.2</figcaption></figure><p>Research usually starts with good intentions. You gather articles, reports, videos, interviews, and notes; you try to make sense of the subject; and then you attempt to turn all of that into something useful — an article, a presentation, a website, or even an automated agent that keeps you posted on new developments. In practice, the process gets messy fast. Key facts get buried in a dozen browser tabs, sources contradict each other, notes pile up without structure, and turning raw research into a finished product becomes its own project.</p><p>One effective way to tame this is to split the work between two AI tools that are good at different things. Google’s Gemini Notebook — the tool most people still know by its recent former name, NotebookLM — is built to work with a defined collection of sources you provide, and to help you understand what they say. Perplexity Computer is built to take that understanding and turn it into something you can use: a finished document, a working website, a connected workflow, or an AI agent that operates on a schedule. Gemini Notebook is the research brain; Perplexity Computer is the production hands.</p><p>This article is not about how to use Gemini Notebook on its own — many guides already cover that ground. It’s about the handoff: what to do with your research once you’ve done it, and specifically, the things you can only accomplish once you move from a closed research notebook into a tool that can act — build files, connect to your calendar and inbox, publish a live page, or keep working in the background on its own.</p><h3>Why the Handoff Matters</h3><p>Gemini Notebook is excellent at grounded understanding. Feed it PDFs, Google Docs and Slides, spreadsheets, web pages, YouTube transcripts, audio recordings, and your own notes, and it will summarize them, compare their arguments, flag disagreements, and answer questions strictly from that material rather than from everything an AI model might otherwise know off the top of its head. As of its mid-2026 rename, it also gained the ability to run code inside a secure notebook environment, so it can now do real data analysis on the sources you have supplied. That is a meaningful upgrade, but it’s still analysis contained inside the notebook: the output is understanding, not a finished deliverable, and the notebook has no way to reach outside itself.</p><p>That is where Perplexity Computer earns its place in the workflow. It doesn’t just answer a question — it can plan and carry out a multi-step project: writing and formatting a document, building a slide deck or spreadsheet, designing and deploying an actual website, drafting outreach that’s ready to send, or specifying and running an AI agent that checks a topic on its own schedule and reports back. It can also reach into the tools you already use — email, calendar, cloud storage, team chat, a CRM — and either read from them for context or, with your approval, act on your behalf. None of that is something a research notebook, however capable, is designed to do.</p><h3>A Simple Workflow</h3><p>The workflow that works best keeps its two halves clearly separated. Start by deciding what you want at the end, not “research artificial intelligence,” but something concrete, like a beginner-friendly article on home solar power or a presentation comparing three project-management tools for a nontechnical audience. The outcome you choose determines everything downstream: which sources matter, how much research is enough, who the audience is, and what format the finished piece should take.</p><p>With that outcome in mind, build a notebook in Gemini Notebook and add material that is authoritative, current, and representative of more than one point of view. Resist the urge to dump in everything remotely related — a handful of well-chosen sources beats hundreds of loosely relevant ones. Then interrogate the material before asking for anything final: have the notebook summarize each source, compare claims, identify where sources agree or disagree, and flag what is uncertain or unverified. This step is what keeps the eventual output honest, separating established fact from interpretation, opinion, and speculation before any of it reaches a draft.</p><p>Once you have that grounded picture, distill it into a short brief: the central question, key findings with their citations, definitions of important terms, open disagreements, and what still needs verification. This brief, not the raw notebook, is what you hand to Perplexity Computer. Treat it as the authoritative source of facts, and be explicit about what Perplexity Computer is and isn’t allowed to do with it: whether it can research the live web to fill gaps or check for updates, how it should label anything it finds that wasn’t in the original notebook, and what actions — sending a message, publishing a page, deploying an agent — require your sign-off before they happen. For anything substantial, ask for a plan before asking for the finished product; reviewing an outline is far cheaper than redoing a finished draft.</p><h3>What Only Perplexity Computer Can Do From Here</h3><p>A few things become possible only once research moves into an environment built for execution rather than analysis.</p><p>The most obvious is producing a genuinely finished deliverable in whatever format you actually need: a formatted document, a slide deck with charts built from your verified data and a references slide, a spreadsheet, or a polished article ready to publish, rather than a chat answer you still have to assemble yourself. Perplexity Computer can take the same brief and adapt it into several of these formats in one pass, which is useful when a single research effort needs to become an article, a set of slides, and a handful of social posts all at once.</p><p>Beyond documents, it can build and deploy a website: an explainer page, an interactive comparison tool, a timeline, or a small dashboard, complete with working charts and navigation, published at a real address rather than only described in the abstract. It can also do research the notebook can’t: running live web searches to check whether facts in your brief have changed, pulling current prices or news, and clearly labeling what came from the original notebook versus what it found afterward, so nothing quietly gets swapped in as though it had always been there.</p><p>Perhaps most distinctively, it can connect to the tools you actually use day to day — email, calendar, cloud storage, team chat, a CRM — and either pull in current context, such as an upcoming meeting or a recent email thread, or, with explicit approval at each step, take action: drafting outreach, updating a record, scheduling a follow-up. And it can go a step further than a one-time document by specifying and running an actual AI agent: a process that monitors a topic on a recurring schedule, compares new material against your existing knowledge base, flags disagreements, and sends you a summary without you having to ask each time. None of this is within reach of a notebook confined to the sources you loaded into it, however good its analysis of those sources may be.</p><h3>Where This Shows Up in Practice</h3><p>The pattern repeats across quite different projects. A competitive-intelligence effort might start with a notebook full of competitor pages, annual reports, and reviews, refined into a brief about positioning and market gaps — but the deliverable that actually gets used, such as a sales battlecard or a market map, and the recurring monitoring agent that watches for the next product announcement, only exist because a production tool built them and can keep them running. Meeting preparation follows the same shape: the notebook can profile the people involved and surface open issues from past correspondence, but pulling in an actual upcoming calendar event, checking for a company’s latest announcement, and assembling a one-page briefing that’s ready to bring into the room requires a tool that can look at your calendar and act on what it learns. Customer outreach works similarly — a notebook can help you understand a prospect’s priorities from their public materials, but drafting a personalized email and, after your approval, sending it or logging it in a CRM is production work, not research.</p><p>The same logic applies closer to home. Someone organizing letters, photographs, and family interviews into a family history can use a notebook to make sense of the material, but turning it into a shareable booklet, a chronological timeline, or a small searchable archive site that relatives can browse on their own is, again, something only a production tool builds. The research stays the foundation; the finished thing people hold, read, or click through comes from the handoff.</p><h3>Keeping the Handoff Honest</h3><p>The value of splitting the work this way disappears if the boundary between the two tools gets blurry. Be explicit with Perplexity Computer about what’s authoritative, whether it’s allowed to bring in new information from the open web and how that should be labeled, which claims need citations, and — critically — which actions need your explicit go-ahead before they happen: sending anything, publishing anything, spending money, changing records, or leaving an agent running with standing permissions. Review the work in stages rather than only at the end. Check that the sources were sound before trusting the brief built from them, check that the brief accurately reflects the sources before trusting a draft built from it, and check the finished draft or agent before approving anything irreversible. AI can compress the time between a pile of raw material and a finished, usable result, but the judgment about what is true, what is ready, and what is safe to release still belongs to the person running the process.</p><h3>The Bigger Picture</h3><p>The real gain here is not asking one AI system to do everything — it is letting each tool do what it is good at. Gemini Notebook turns a disorganized collection of documents, recordings, and notes into a grounded understanding of a subject. Perplexity Computer turns that understanding into an article, a presentation, a deployed website, a connected workflow, or a carefully bounded AI agent — the parts of the job a research notebook was never built to do. Research carefully, structure what you have learned, hand it off with clear rules, and let the production side build. Just keep a human in the loop before anything goes out the door.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0dbb7a13a143" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Nano Banana Editing Power]]></title>
            <link>https://medium.com/@terry.faircloth/nano-banana-editing-power-34c69e73066e?source=rss-3a2f1afb56a4------2</link>
            <guid isPermaLink="false">https://medium.com/p/34c69e73066e</guid>
            <category><![CDATA[aiimages]]></category>
            <category><![CDATA[how-to]]></category>
            <category><![CDATA[nano-banana]]></category>
            <category><![CDATA[image-editing]]></category>
            <category><![CDATA[editing-prompts]]></category>
            <dc:creator><![CDATA[Terence Faircloth]]></dc:creator>
            <pubDate>Sun, 08 Mar 2026 02:17:23 GMT</pubDate>
            <atom:updated>2026-03-08T15:44:20.834Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*04izxast6dKCdsK1U1JtUA.jpeg" /><figcaption>AI image created by the author in Midjourney 7</figcaption></figure><p>Nano Banana makes creative photo editing feel playful and approachable, even if you’ve never touched pro software before. Instead of wrestling with layers and sliders, you simply tell it what you want — “turn this cloudy sky into a sunset,” “give this portrait a moody film look,” or “make this street scene feel like a watercolor painting” — and it reimagines the image for you. It’s just as comfortable cleaning up distractions as it is adding entirely new elements, like swapping backgrounds, changing outfits, or shifting the mood of a scene from bright and cheerful to dark and cinematic. Because it understands the overall structure of the image, it usually keeps faces, poses, and key details intact while experimenting with color, light, and style, so you can push your ideas further without breaking the photo.</p><p>I wanted to make an image of a Rock Band, so I created the image shown above using Midjourney 7. The image did not catch what I was trying to do, so let’s see how it can be edited with Nano Banana. I use Google AI Pro which has Nano Banana included in my subscription. Here’s how it works.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/767/1*8Bheuajr0Eo8O4gX4Cj4kQ.png" /><figcaption>Prompts are entered in the dialog box, and images are uploaded using the + and a pull down menu.</figcaption></figure><p>Any Rock Band should have a vocalist, so let’s add a vocalist to the band.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*C46GZOlT2w9zwZNUSXCPsg.png" /><figcaption>Prompt — Add a vocalist with microphone</figcaption></figure><p>The microphone by the drummer seems to be floating in mid air; let’s get rid of it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bspGnYa6GnMIB1vTn3N5QA.png" /><figcaption>Prompt — Remove the microphone at the upper right</figcaption></figure><p>I think blue and red tones are overpowering the image, and I would like greater detail of the musicians.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-M-W6-G7PvgOdeN3deKLxg.png" /><figcaption>Prompt — Transform the image into cinematic Instagram aesthetic photography. Balanced color grading, soft warm tones, smooth skin texture, cinematic lighting, subtle film grain, dreamy background blur, stylish highlights and shadows, editorial lifestyle photography mood, ultra realistic details, modern aesthetic photo style.</figcaption></figure><p>The Man in the middle is too close to the vocalist, so let’s give them more space.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UkccQP5WyOGvOOEXiPjUzg.png" /><figcaption>Prompt — Move the man in the middle back from the vocalist</figcaption></figure><p>The instruments are wrong; there should be a lead guitar and a four string bass electric guitar. I had trouble getting Nano Banana to understand what I wanted, so I found an image of a four string bass electric guitar.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*enmP4MXWnARD8btn-N3weg.jpeg" /><figcaption>creative commons image of a bass guitar</figcaption></figure><p>Now we need to have Nano Banana incorporate the four string bass electric guitar into the image. I uploaded the bass guitar image file by clicking the + sign and selecting the file on my computer using the the prompt box in Nano Banana, then my prompt directed Nano Banana to use the uploaded image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M5byC9TkDQiGvzm2AALk3A.png" /><figcaption>Prompt — Replace the guitar played by the man in the middle with a four string electric bass guitar per the uploaded image</figcaption></figure><p>That worked, but somehow Nano Banana changed the aspect ratio of the image to a square, so let’s fix that.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OhLSoUAK-98imOi3WiNU3w.png" /><figcaption>Prompt — change the aspect ratio to 2380x1792</figcaption></figure><p>And finally, let’s change the guitar on the left to a Fender Stratocaster.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LHz-j3VHxJw9s0_WLPCDkQ.png" /><figcaption>Prompt — change the guitar on the left to a Fender Stratocaster, the neck of the guitar should be behind the vocalist.</figcaption></figure><p>Voila, we’ve got our rock band with a vocalist, lead guitar, electric bass guitar, and drums; we’ve also changed the mood and added detail. This is how simple it is to edit in Nano Banana.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=34c69e73066e" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>