<?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 Tuba Tezer on Medium]]></title>
        <description><![CDATA[Stories by Tuba Tezer on Medium]]></description>
        <link>https://medium.com/@tubatezer?source=rss-8487a91a2673------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*ahJ_GCN_Ew0Bk-NE00ovcQ.png</url>
            <title>Stories by Tuba Tezer on Medium</title>
            <link>https://medium.com/@tubatezer?source=rss-8487a91a2673------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 17:04:37 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@tubatezer/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[Bye Medium, Hi Substack]]></title>
            <link>https://medium.com/@tubatezer/bye-medium-hi-substack-83bf7fe7c60a?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/83bf7fe7c60a</guid>
            <category><![CDATA[product-blog]]></category>
            <category><![CDATA[substack-vs-medium]]></category>
            <category><![CDATA[ai-in-product-development]]></category>
            <category><![CDATA[substack]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Sun, 28 Jun 2026 10:18:05 GMT</pubDate>
            <atom:updated>2026-06-28T10:18:05.468Z</atom:updated>
            <content:encoded><![CDATA[<p>I dusted off my Medium account recently to publish something new, and realized something a little sad and a little freeing: nobody’s really here anymore.</p><p>Medium was good to me. It’s where I first started writing in public about AI, organizations, and the messy gap between strategy and reality. But the energy has moved, and after poking around, it’s pretty clear the conversation I want to be part of now lives on Substack.</p><p>So, I’ve moved.</p><p>The new home is called <a href="https://runsonvibes.substack.com/"><strong>Runs on Vibes</strong>.</a> It’s a single-lens newsletter about the things I can’t stop thinking about: how AI actually takes root inside organizations (and why it so often doesn’t), product and building, navigating a tech career in Germany, and the books and ideas that reshape how I see all of it. Strategy, plainly stated, with an actual point of view.</p><p>The first post is already up. It’s about why most AI rollouts inside companies aren’t really failing, they’re quietly repricing who’s valuable, and almost nobody names it out loud.</p><p>If you’ve enjoyed anything I’ve written here, come along. You can find Runs on Vibes here: <a href="https://runsonvibes.substack.com/">https://runsonvibes.substack.com/</a></p><p>I’ll leave the old posts up. But the new stuff, and the better stuff, is over there now.</p><p>See you on the other side.</p><p>Best,<br>Tuba</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=83bf7fe7c60a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your Org Runs on Vibes. AI Agents Won’t.]]></title>
            <link>https://medium.com/swlh/your-org-runs-on-vibes-ai-agents-wont-40b958e76dc2?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/40b958e76dc2</guid>
            <category><![CDATA[conversational-ai]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[organizational-design]]></category>
            <category><![CDATA[decision-making]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Sun, 31 May 2026 13:26:09 GMT</pubDate>
            <atom:updated>2026-06-06T19:06:00.477Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ceQpPzyhkh4LfAPTU9d2RQ.png" /></figure><p>A lot of organizations run on what you might call latent state.</p><p>The org chart says one thing. The actual decision-making process is something else entirely.</p><p>Humans navigate this surprisingly well. We absorb context continuously. We know who usually gets the final say, whose opinion carries extra weight, which decisions require consensus, and which ones only appear to. Most of this information is never written down. It lives in conversations, relationships, history, and intuition.</p><p>It mostly survives, because humans are forgiving systems. When accountability is fuzzy, we improvise: read the room, defer to the senior voice, argue it out, muddle through. The cost is real but absorbable: slower decisions, recurring debates, the occasional turf standoff.</p><p>In 1970, Jo Freeman called this the <em>tyranny of structurelessness</em>: refuse to formalize structure and you don’t remove power, you just make it informal and unaccountable, held by whoever has the most tenure, confidence, or persistence.</p><p>As teams get smaller and faster in the AI era, the cost of unclear decision rights goes <em>up</em>, not down. Structurelessness scales worse the leaner you run.</p><h3>What changes when a non-human starts deciding</h3><p>An AI agent doesn’t join a company and spend six months learning the social topology. It sees only the context you provide and the actions you permit. If an organization relies on unwritten conventions to function, the agent has no access to them.</p><p>This creates an interesting effect.</p><p>Many companies believe they are decentralized and autonomous. In reality, they are running on a dense layer of human inference.</p><p>Agents remove that compensation layer.</p><p>Drop an agent that issues refunds, ships copy, reprioritizes a backlog, or makes a product call into a team that never defined who is allowed to decide what, and you haven’t added a helpful teammate. You’ve handed real, fast, scaled power to an actor inside a system that has no way to say who owns the result.</p><p>The question stops being “who decides among us” and becomes <strong>who’s accountable when the agent decides.</strong> A team that struggles to answer the first version will find the second much harder.</p><h3>Conversational AI has been quietly solving this for years</h3><p>I’ve spent years building conversational AI bots, and now agents in production, handling thousands of customer conversations a day across dozens of countries. Here’s the thing difficult to realize for those outside that world: the hard part was never the model. The hard part was governance.</p><p>Consider a simple question: can an agent issue a refund?</p><p>This sounds like a model capability question. It usually isn’t.</p><p>The difficult part is not generating the response. The difficult part is specifying the authority.</p><p>Can it approve, or directly escalate?<br>Can it make exceptions?<br>Who owns the outcome?</p><p>The model is often the easiest component in the system. The hard part is defining the operating environment around it.</p><p>A useful agent is rarely given unlimited authority. Instead, it operates within carefully defined boundaries. Some decisions it owns. Others it escalates. The quality of the system depends as much on these boundaries as on the intelligence itself.</p><p>In that sense, agent design starts to look surprisingly similar to organizational design.</p><h4>Conversational AI teams have already spent years solving the problem of assigning decision rights to non-human actors, and product organizations are about to rediscover the same discipline.</h4><p>In other words: conversational AI teams have been designing decision rights for non-human actors for years. Define precisely what the agent owns, what it must hand back, and who’s accountable for each. That discipline (unglamorous, all about boundaries) is exactly what every product org is about to need. Most have never built it even for their humans.</p><h3>The fix isn’t more structure. It’s firmer structure.</h3><p>The fix isn’t more process. It’s a better specification. Every effective system has clear interfaces. Organizations are no different.</p><p>When a decision matters, three things need to be explicit:</p><p><strong>Who owns the decision.</strong> Not who has input. Not who has strong opinions.</p><p><strong>Who owns the outcome.</strong> Authority and accountability need to terminate in the same place. When they don’t, decisions get optimized for consensus instead of results.</p><p><strong>What success looks like.</strong> If teams don’t share a definition of “good”, every disagreement becomes a debate about taste, politics, or seniority.</p><p>This sounds like management advice. It’s really an interface design problem.</p><p>An agent only sees the interface you expose. If ownership, authority, and success criteria are implicit, the agent inherits the ambiguity and scales it.</p><p>An organization is, among other things, a mechanism for allocating authority.</p><p><strong>Humans can function even when that allocation is ambiguous. Agents force the ambiguity into the open.</strong></p><p>One way to think about the next few years is that organizations themselves are becoming part of the software stack.</p><p>Not because companies are turning into software, but because assumptions that previously lived in people’s heads now need to be specified explicitly. Decision rights, escalation paths, ownership boundaries, success criteria. These become operational requirements.</p><p>The interesting possibility is that the bottleneck to deploying capable agents may not be the agents themselves. It may be that organizations are less well specified than we thought.</p><p>The teams that do well with AI agents won’t be the ones that eliminate decision-making structure. They’ll be the ones that make it visible and honest about what Freeman saw fifty years ago: structure exists whether you design it or not. The only real choice is whether it’s explicit and accountable or invisible and making decisions anyway.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40b958e76dc2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/your-org-runs-on-vibes-ai-agents-wont-40b958e76dc2">Your Org Runs on Vibes. AI Agents Won’t.</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What is a Flash Briefing, and How Do You Make One?]]></title>
            <link>https://medium.com/swlh/what-is-a-flash-briefing-and-how-do-you-make-one-b45af5cf9c4c?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/b45af5cf9c4c</guid>
            <category><![CDATA[amazon-echo]]></category>
            <category><![CDATA[google-assistant]]></category>
            <category><![CDATA[google-home]]></category>
            <category><![CDATA[alexa-skills]]></category>
            <category><![CDATA[amazon-alexa]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Fri, 09 Nov 2018 15:12:11 GMT</pubDate>
            <atom:updated>2018-11-13T13:10:26.159Z</atom:updated>
            <content:encoded><![CDATA[<p>Making an Alexa skill might sound tough — there are so many queries and responses to account for! But many brands are taking a simpler approach to conversational voice interfaces: Alexa flash briefings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/550/0*68N5ec32INvSFCq9.png" /><figcaption>Source: developer.amazon.com</figcaption></figure><p>First, what is a flash briefing? <strong>Flash briefings are short, informative pieces of prerecorded audio </strong>— kind of like a podcast or radio show.</p><blockquote>“Alexa flash briefings can last up to ten minutes and are usually pushed out on a daily or weekly basis.”</blockquote><p>Because they’re short and broadcast regularly, Alexa flash briefings make for a great content channel for brands. Some things you can do with an Alexa flash briefing include informing your fans of updates on products, services or news relating to your brand.</p><blockquote>TIP 1: You can also provide short-form lifestyle content like a daily horoscope or word-of-the-day.</blockquote><blockquote>TIP 2: Influencers can also set up an Alexa or Google Home news briefing to provide professional advice or daily tips to their audience.</blockquote><p>Think of flash briefings as a cross between podcasts and Alexa skills. They’re not as interactive as skills, but allow Alexa users to receive all the news they care about most within a conversational interface.</p><h3><strong>Alexa Flash Briefing Tutorial</strong></h3><p>Setting up an Alexa flash briefing is a relatively simple process. First, you’ll need a place to host your audio content. While you can self-host, popular services for hosting audio content include <a href="https://www.libsyn.com/">Libsyn</a> and <a href="https://effct.co/">Effct</a>, the latter of which is free.</p><p>No matter where you choose to host, you’ll need to copy the RSS feed generated for your flash briefing. If you choose to develop a Google Home news briefing, you’ll need to do a little extra work with your RSS feed — but we’ll get to that below. If you’re developing briefings for Amazon, move on.</p><p>The next step in our Alexa flash briefing tutorial is to set up an Alexa skill for delivering the briefing. First, visit the Amazon Developer Console to create a new skill. Begin by giving it a name, then click “next.” You’ll be offered a selection of Alexa skill models to choose from. Select “flash briefing” from the list, then you’ll be met with a series of Alexa flash briefing options.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Wp6VgqUrpIbpj73MK1mkRA.png" /><figcaption>Amazon Developer Console</figcaption></figure><p>The first of the Alexa flash briefing options to include is an error message. Alexa will read this to your audience if your briefing is unavailable. Next, you’ll be asked to paste the RSS feed provided by your audio host. The “preamble” field is a short, introductory cue that Alexa will read before delivering your briefing. Finally, fill out any remaining Alexa flash briefing options — frequency, briefing name, category/genre and more — and soon you’ll be ready to launch.</p><p>Once you’ve filled out the page, it’s time to make a listing or profile page for your flash briefing. This process is the same as with any Alexa skill: give your briefing a short description, a long one and a profile image to use. After that, submit your skill and begin recording your content! When you upload new content to your audio host, it will automatically push to your flash briefing via RSS.</p><h3><strong>Publishing a Google Home News Briefing</strong></h3><p>Google Assistant doesn’t have flash briefings in the Alexa sense — instead, you’ll be publishing your audio content as a podcast. This is a little more technical than Alexa’s process. First, your briefing will need its own homepage. Second, you’ll need to edit the briefing’s RSS feed to include snippets of code that are required for Google Assistant to recognize it in its directory — check out all the requirements <a href="https://developers.google.com/actions/content-actions/podcasts#rss-feed-requirements">here</a>. Google doesn’t require setting up an Assistant action. Once you’ve included the necessary code in your RSS feed, your podcast will show up automatically within search results.</p><h3><strong>Next Steps: Set Up an Interactive Skill</strong></h3><p>Flash briefings are great for quickly informing your audience, but a fully interactive Alexa skill or Google Assistant action will allow even more freedom. With a well-designed skill, you can provide personalized service and content to your audience at any time — not just once a day or week. With a <a href="https://botanalytics.co/voice-analytics">voice chatbot analytics tool</a> like we provide as Botanalytics, you can also receive helpful usage data to get to know your audience even better.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*olgRlUo4hl0Jt2s_VoJmTw.png" /><figcaption>Botanalytics session flow for voice apps</figcaption></figure><p>Flash briefings make for a great, simple way to dip your toes into conversational platforms. Once you’ve cultivated a presence on the platform, try implementing a fully featured skill next!</p><figure><a href="https://medium.com/swlh"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YqDjlKFwScoQYQ62DWEdig.png" /></a></figure><h4>This story is published in <a href="https://medium.com/swlh">The Startup</a>, Medium’s largest entrepreneurship publication followed by +388,268 people.</h4><h4>Subscribe to receive <a href="http://growthsupply.com/the-startup-newsletter/">our top stories here</a>.</h4><figure><a href="https://medium.com/swlh"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ouK9XR4xuNWtCes-TIUNAw.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b45af5cf9c4c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/what-is-a-flash-briefing-and-how-do-you-make-one-b45af5cf9c4c">What is a Flash Briefing, and How Do You Make One?</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[These Alexa Skill Platforms Let You Build an App in Just Minutes]]></title>
            <link>https://medium.com/@tubatezer/these-alexa-skill-platforms-let-you-build-an-app-in-just-minutes-c63bfc4823c2?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/c63bfc4823c2</guid>
            <category><![CDATA[voice-summit]]></category>
            <category><![CDATA[voice-app]]></category>
            <category><![CDATA[amazon-alexa]]></category>
            <category><![CDATA[voice-assistant]]></category>
            <category><![CDATA[alexa-skills]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Fri, 03 Aug 2018 22:50:14 GMT</pubDate>
            <atom:updated>2018-08-09T19:05:40.135Z</atom:updated>
            <content:encoded><![CDATA[<p>The voice app market is booming; <a href="https://medium.com/u/56f4af0378c1">Dave Isbitski</a> from Amazon Alexa announced two weeks ago during Voice Summit in NJIT that they have more than 45,000 skills in Alexa (It was around 26k when I checked 2,5 months ago, huge increase). As of this spring, more than <a href="https://techcrunch.com/2018/03/07/47-3-million-u-s-adults-have-access-to-a-smart-speaker-report-says/">47 million US adults</a> are using voice-activated smart speakers powered by those skills. With the ever-growing popularity of voice apps with consumers, there’s no better time for Alexa skill developers or those who want to build a Google Home action.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nO2iT0v17DIgVfWjpsV3LA.jpeg" /><figcaption>Voice Summit, Newark 2018</figcaption></figure><p>And thankfully, developing such a skill doesn’t need to be difficult. While it might seem complicated or intimidating to build a voice-controlled AI assistant, even a layperson can get a skill up and running in a matter of minutes. With an easy-to-use Alexa skills platform and <a href="https://botanalytics.co/voice-analytics">analytics for voice bots</a> that we provide as <a href="https://botanalytics.co/">Botanalytics.co</a>, publishing a great voice app is a snap — here’s my picks for some of the best building platforms.</p><p><a href="https://botsociety.io/"><strong>Botsociety</strong></a></p><p>Botsociety helps you design conversational interfaces for voice and chat. Do you have a story to tell? This tool helps you design and tell your story in a very simple way. You can also invite your team and work collaboratively. They have a free starting package with all the features included, if you want to work on more than one project you need to check other subscription packages starting from $49/month. Novice Alexa skill developers can easily create a simple skill in just minutes with the simple visual editor.</p><p><a href="https://getstoryline.com/"><strong>Storyline</strong></a></p><p>Storyline is a fast, simple Alexa skill builder. This Alexa skills platform doesn’t require any coding; just start with a welcome message, then add some responses in a simple flowchart interface. Storyline is ideal for choose-your-own-adventure style games, simple conversations or news/media briefings. Most of Storyline’s features are free, though those looking for a beefier Alexa skills platform can sign up for an enterprise-ready subscription at $49/month.</p><p><a href="https://voiceapps.com/"><strong>Voice Apps</strong></a></p><p>With a simple name comes a simple Alexa skill builder. Just like Storyline, Voice Apps uses an easy drag-and-drop editor that pretty much anyone can use. So why should one use Voice Apps instead of other tools? One of Voice Apps’ strongest features is its sample skills, which lets Alexa skill developers immediately create skills by customizing readymade intents and responses. Again, this is an ideal Alexa skills platform for novices or experienced users alike. The basic plan is free, but those in need of some added features will only have to pay $19.95/month for the full suite.</p><p><a href="https://blu.ai/"><strong>Blutag</strong></a></p><p>Blutag is multiplatform, which means you can use if as a Google Home developer platform in addition to making Alexa skills. Its codeless interface, just like those above, makes it beginner-friendly. Just like Voice Apps, this Alexa skill builder offers a variety of skill templates to get a head start on building useful voice applications. Blutag is particularly great for ecommerce brands dipping their toes into Alexa skills, since it offers simple integration into existing ecommerce and retail platforms.</p><p><a href="https://www.jovo.tech/"><strong>Jovo</strong></a></p><p>Okay, Jovo is a bit different than any Alexa skills platform or Google Home developer platform detailed above. This one <em>does</em> require coding, although it does tout itself as being user-friendly thanks to its relatively simple framework. It also comes with templates, so even if coding knowledge is limited, beginners can experiment with premade skills when developing their own.</p><p>If you’re an experienced coder, you might wonder: what’s the <em>real</em> benefit to using Jovo? The platform is surprisingly community-focused. First, it’s totally free and open source to use. Second, Jovo offers a Slack channel for Google and Alexa skill developers using the platform, so you can bounce off ideas or get help from other users if needed. That in itself might make it a great environment for new developers, even if they aren’t so skilled with code.</p><p><a href="https://conversation.one/"><strong>Conversation.One</strong></a></p><p>Conversation.One is, once again, an excellent choice for developing Google Home actions and Alexa Skills alike thanks to its visual editor. It also offers both a free and a pay-as-you-go plan, making it an affordable option for brand-new skills or those that need more support while scaling.</p><p>What sets Conversation.One apart from other platforms, though, is that it can be used to make voice apps as well as text-based chatbots. If you’re developing an application that can be used via voice as well as a text-based platform like Facebook Messenger, this would be an easy platform for implementing it.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c63bfc4823c2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Next Big Thing? Chatbots]]></title>
            <link>https://medium.com/swlh/the-next-big-thing-in-content-marketing-chatbots-54ad20603c25?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/54ad20603c25</guid>
            <category><![CDATA[chatbots]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Tue, 24 Jul 2018 11:23:47 GMT</pubDate>
            <atom:updated>2020-07-08T09:55:35.399Z</atom:updated>
            <content:encoded><![CDATA[<p>We’ve all been helped by the customer service chatbot or assistant dedicated to a specific task. But businesses who really want to make a splash in the chatbot space are turning their attention to chatbots as content delivery systems. Why should businesses focus on content in chatbots? <a href="http://www.businessinsider.com/the-messaging-app-report-2015-11">With users focused on messaging instead of their social feeds</a>, messaging platforms have the potential to become the next big channel for pushing content and marketing. This means brands hoping to stay relevant in their industries must be ready to adapt.</p><h3><strong>Why Use Chatbots for Content?</strong></h3><p>One of chatbots’ greatest features is that te are excellent for user personalization. While content traditionally revolved around search terms or trends for discoverability and relevancy, chatbots can provide users with relevant content through the vast amounts of data they get from the user.</p><blockquote>“More personalized delivery makes your content more irresistible to users, helping to boost CTR and user satisfaction.”</blockquote><p>Next, chatbots-as-content help you humanize your brand and connect with users in a casual way that feels natural to them. For example, you might create a friendly, funny character who serves as a brand ambassador. That said, simply having a good personality isn’t enough for a <a href="https://botanalytics.co/blog/2017/10/02/develop-chatbot-content-strategy/">chatbot content strategy</a>, though we’ll delve into how you can build excellent content chatbots below.</p><h3><strong>Design a Character Around User Need</strong></h3><blockquote>“An effective content chatbot should find a novel way to support your brand or company’s goals and values — just like your existing product or service does.”</blockquote><p>What problem does your brand aim to solve, and how can a chatbot support it? One amazing example is Casper’s <a href="http://insomnobot3000.com/">Insomniabot-3000</a>. Casper is a mattress brand that aims to give customers a great night of sleep, but those with insomnia will have trouble sleeping no matter their bed. Insomniabot-3000 makes things a little better by giving insomniacs someone to talk to late at night when their friends are all asleep, supporting the brand’s mission — helping those with low-quality sleep — in a creative and humorous way.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/984/1*lPcfm35m39xEgrOpCipcuA.png" /><figcaption>Casper’s Insomniabot-3000</figcaption></figure><h3><strong>Augment the Experience You Already Deliver</strong></h3><p>This content strategy is perfect for brands in the entertainment industry. A content chatbot can bring an imagined world to life, allowing the user to put themselves within it. This can be done through discussions with a mascot or fictional character or through a game-like experience. One company that has pulled off both styles well is Activision, who has used chatbots for augmenting its popular <em>Call of Duty</em> videogame series.</p><p>The company’s first <em>Call of Duty</em> chatbot is one <a href="https://venturebeat.com/2016/05/03/call-of-duty-infinite-warfares-first-victory-6m-bot-messages-on-facebook/">based on Lt. Reyes</a>, the lead character of <em>Call of Duty: Infinite Warfare</em>. Activision made a spectacle of the chatbot by hiding Easter eggs within the previous game, which directed users to the bot on Facebook. The chatbot received over 6 million messages on its first day, which directed users to an exclusive trailer for the upcoming game. Activision could have released the trailer and other content in a traditional way, but their Lt. Reyes chatbot instead blurred the boundaries between fiction and reality, providing a unique way for dedicated fans to engage with the world they had built.</p><p>Activision’s second chatbot in the series, which released in the week leading up to <em>Infinite Warfare’s</em> release, took the form of a <a href="http://www.alistdaily.com/social/call-duty-infinite-warfare-explores-space-facebook-messenger/">guided tour</a> via a choose-your-own-adventure style experience. Users chat with Alana, an in-universe tour guide, who directs them through the game’s futuristic take on the solar system with humorous results. This experience, different than the first in character and tone but used to augment and promote the same product, shows how you can refresh an experience and re-engage a userbase with chatbots.</p><h3><strong>Leverage Existing Content in a New Way</strong></h3><p>How do you ensure your evergreen content doesn’t get lost in the aether after it’s been promoted online? While you could rely on SEO alone, a solid alternative would be to push it to a user via a chatbot whenever it’s most relevant in conversation.</p><p>Because chatbots offer sophisticated personalization and instantaneous communication at any time, they’re a great way to help users discover content that might otherwise go unnoticed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/374/1*79Couc0gUjbZQx-QAgGKOA.png" /></figure><p>The important thing to remember here is <em>when, where, why and how</em> users will want or need your content. A wonderful example is the <a href="https://www.foodnetwork.com/site/apps/chatbot">Food Network chatbot</a>, which lets users easily search for and discover recipes form the Food Network website in several easy ways. Its most impressive feature is that users can ask for recipes recently featured on TV (pairing well with users’ viewing habits) or call up ingredient-based recipes while in the kitchen hands-free (via the Alexa skill).</p><h3><strong>Keep Up with Chatbot Analytics to See What’s Working</strong></h3><p>How do you know how effective your content chatbot is at engaging users? Just like tracking analytics for web content, you’ll want to keep up with <a href="https://botanalytics.co/">conversational analytics</a> to see what’s engaging users and what’s giving them problems. For example, by tracking the most-used phrases or words in your conversations, you can discover what topics matter most to your users — and when. From there, you can more effectively design your content strategy across all platforms.</p><figure><a href="https://medium.com/swlh"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YqDjlKFwScoQYQ62DWEdig.png" /></a></figure><h4>This story is published in <a href="https://medium.com/swlh">The Startup</a>, Medium’s largest entrepreneurship publication followed by 351,974+ people.</h4><h4>Subscribe to receive <a href="http://growthsupply.com/the-startup-newsletter/">our top stories here</a>.</h4><figure><a href="https://medium.com/swlh"><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ouK9XR4xuNWtCes-TIUNAw.png" /></a></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=54ad20603c25" width="1" height="1" alt=""><hr><p><a href="https://medium.com/swlh/the-next-big-thing-in-content-marketing-chatbots-54ad20603c25">The Next Big Thing? Chatbots</a> was originally published in <a href="https://medium.com/swlh">The Startup</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Will the AI Overlords Use Our Data for Evil? Probably Not]]></title>
            <link>https://medium.com/chat-bots/will-the-ai-overlords-use-our-data-for-evil-probably-not-5013619e8dfc?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/5013619e8dfc</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[gdpr]]></category>
            <category><![CDATA[data-security]]></category>
            <category><![CDATA[bots]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Thu, 24 May 2018 18:05:47 GMT</pubDate>
            <atom:updated>2018-05-25T00:10:35.911Z</atom:updated>
            <content:encoded><![CDATA[<p>It’s almost like something from a dystopian movie: the powers-that-be using vast amounts of surveillance, an all-seeing eye, to manipulate and control the minds of the masses. This is how lots of people have felt upon learning that Cambridge Analytica, a British political consulting firm, mishandled Facebook user data.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/593/1*bdYuSejamWm-xRnoM7Sgcw.png" /></figure><p>After the Cambridge Analytica scandal, many users are concerned with how their data is handled by online services. And rightly so; our data is collected from almost every corner of the web, and it can be difficult to understand how it’s used.</p><p>It’s enough to make someone want to quit the internet altogether — but don’t start packing a suitcase to live off-the-grid just yet.</p><p>While Facebook has come under fire in the past 2 months or so, it’s important to understand that the company isn’t alone in harvesting vast amounts of data from users. It’s also important to know that the problem isn’t the fact that data was collected; the issue is how Cambridge Analytica <em>got ahold of</em> the information and used it.</p><p>It definitely sounds creepy that Facebook, Google and many other websites and services you use collect data. No one wants a super-nosy friend prying into all their business. But this data is routinely used to provide better services to you: Facebook can suggest content relevant to your interests, Google can surface up results to a query based on your lifestyle, and chatbots can get to know you just like a friend based on the information you’ve given it.</p><p>That’s all well and good, you tell me. But how can we make sure the people we give our data to are actually using it for these purposes?</p><p>We should all have a good understanding of what’s being collected, <em>why</em> it’s being collected, and who’s seeing it. This way, we can make informed decisions for protecting our personal data. A lot of businesses are doing a great job at this: when installing a smartphone app, for example, the Android or iOS system will alert you to individual permissions to grant (or block from) the app. And you’ve undoubtedly noticed that websites routinely alert visitors about their cookie policies and why they’re in place, often offering links to easily opt-out.</p><p>While this is simple enough for websites and apps, <strong>what about chatbots?</strong> The technology, whose biggest feature is personalization, must certainly collect and make use of tons of personal data, right? While the vast majority of chatbots make use of personal data to do their jobs well, there’s no need for you to panic! Just like with websites and apps, there are a handful of questions to ask yourself (and a chatbot provider) to gauge trust.</p><h3><strong>What company or individuals are behind the chatbot?</strong></h3><p>This question is important for any chatbot or service, but it’s especially important for more sensitive activities like health or financial services. You wouldn’t just hand over your bank account and routing number to a random stranger over the phone, right? Well, you shouldn’t for a chatbot, either. When Vero, a hot new social network, began trending, many users flocked to the space — until the blogosphere <a href="https://www.thedailybeast.com/as-vero-blows-up-backlash-builds-against-shady-social-app">discovered the politics behind the network’s owner seemed shady at best</a>. If you’re concerned about the data you hand over and how it can be monetized (or used to influence you), first do some research behind the company or individuals offering the service.</p><h3><strong>What security does the business offer?</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*Zq1tnUOhOJRj5amu-nCEmA.jpeg" /></figure><p>If hacking is your concern, know that chatbots use the same HTTPS protocols that websites and apps use. Chances are, that just sounds like a lot of mechanical mumbo-jumbo; all you need to know is that this means your data is pretty secure from data breaches or hacking. If you’re happy to check your bank account online, you should be confident enough against data breaches with chatbots, too.</p><p>If you’re still concerned — it doesn’t hurt to be safe rather than sorry — you may research (or ask) the botmaker about what methods they use to secure user data. It’s also worth checking whether the messaging platform you’re using to connect with the bot is encrypted.</p><h3><strong>What data is collected, and how is it used?</strong></h3><p>Different chatbots collect different types of data to provide better service to you. If you don’t want to reenact “50 First Dates” with a chatbot — having to introduce yourself, your preferences and problems over and over and over — then you’ll have to hand over some data for the chatbot to recognize you and remember your needs. This can include location data as well as times that you’re actively reading or responding to messages.</p><p>Obviously, expect that <em>all</em> messages you send are saved — and could be read by actual humans. Many botmakers will look directly at conversations to understand how their bot is being used and to see how errors may be avoided. This means anything you say can be read and possibly tied to your identity.</p><p>Woah! That sounds scary! But it shouldn’t really be, so long as you are aware. Think about the last time you called a customer service number: the first thing you’re told is that the conversation is being recorded and may be reviewed in the future. This doesn’t stop you from providing the customer service rep with all the information they require from you. In this respect, there’s really no need to be worried about doing the same with a chatbot. That said, don’t be afraid to reach out to a botmaker (or read a privacy policy) with any questions you have about data collection.</p><h3><strong>Are third parties accessing your data?</strong></h3><p>Alright, so you’ve considered all the questions above and might have even reached out to the people who make the chatbot. You’ve decided you trust the company behind it — but is that enough? Chances are, the company that made the chatbot aren’t the only ones who will receive (and possibly make use of) your data. But this isn’t something to panic about! <strong>The botmaker should be transparent with any third parties who will have access to your data.</strong> Here’s what to look for.</p><p>The most common situation in which a third party will hold your data is if the botmaker is storing that data in the cloud. If you’ve used a service like Dropbox or Google Drive, then you’ve already dealt with cloud storage in the past. Many enterprises — chatbots included — use these services to hold and secure their vast amounts of data. Most users don’t need to be concerned about enterprises storing data with these parties, but it <em>is </em>your right to know if this is the case. If you are concerned, check whether the botmaker uses cloud-based services (and which, so you can check <em>their</em> data policies) or if they store data on-premises.</p><p>Most users never stop to consider how the chatbot AI is trained, either. “Training” is exactly what it sounds like: a team of humans training the AI to respond appropriately to queries. This is done based on queries and conversations made with the bot. In short, humans might be reading some of your messages to train the AI. If training is done in-house, you really don’t have much to worry about; it’s a lot like the example above of having a customer service call recorded. But you <em>should</em> be aware of whether training using your data is being done by a third party, which is often the case.</p><p>In some cases, the chatbot company might sell user data to other parties or advertisers. It’s up to you how to feel about this, but one thing is for certain: the company should certainly be transparent if this is a policy of theirs. Again, don’t be afraid to ask the botmaker if you’re ever unsure, or to search through their data policy.</p><p>We at <a href="https://botanalytics.co/">Botanalytics</a> always protect sensitive user data and never share with the third parties.</p><h3><strong>What is the data policy of the messaging platform?</strong></h3><p>Here’s another third party to think about, almost so obvious to be easily taken for granted: your messaging platform. Platforms will also save <em>all</em> communication to ensure their services are running smoothly. Unlike the third parties mentioned above, this one is really out of the botmaker’s control. Whether talking to a chatbot or talking to a close friend, it’s important to know how the messaging platform will use your data (the content of your messages). For example, Facebook’s data policy is a blanket one that covers <em>all</em> its services, including Facebook Messenger. It can be assumed, therefore, that information shared on that channel could be used to help marketers target you for advertising.</p><p>Again, this amounts to user trust. Messaging platforms do a great job at describing what data they collect, why, and how they use it. If you’re already on the platform, chances are you trust it or have an understanding already about its data policies. If you’re uncertain, research that platform’s privacy policy to address any concerns you may have.</p><p>Just like anyone else, chatbots aren’t inherently dangerous when it comes to revealing personal data — but you shouldn’t trust just <em>anybody</em>. With the tips above, you’re better equipped to understand how your data is used and why.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F58521e%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F58521e%2F&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/c90883b0516ddbe5c59960c12a04bc64/href">https://medium.com/media/c90883b0516ddbe5c59960c12a04bc64/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5013619e8dfc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/chat-bots/will-the-ai-overlords-use-our-data-for-evil-probably-not-5013619e8dfc">Will the AI Overlords Use Our Data for Evil? Probably Not</a> was originally published in <a href="https://medium.com/chat-bots">Chatbots Magazine</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cambridge Analytica Has Botmakers Worried — But Should They Be?]]></title>
            <link>https://medium.com/chatbots-journal/cambridge-analytica-has-botmakers-worried-but-should-they-be-f358da3c3e01?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/f358da3c3e01</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[data-protection]]></category>
            <category><![CDATA[facebook]]></category>
            <category><![CDATA[mark-zuckerberg]]></category>
            <category><![CDATA[cambridge-analytica]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Tue, 10 Apr 2018 13:14:12 GMT</pubDate>
            <atom:updated>2018-04-26T08:55:24.873Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Su3qm98Sq3TGTYad2hFQQA.jpeg" /></figure><p>The recent <a href="https://www.nytimes.com/2018/03/19/technology/facebook-cambridge-analytica-explained.html">Cambridge Analytica and Facebook scandal</a> has rocked social media, prompting a great deal of worry and outrage among users concerned about digital security and the safety of their personal data. Even <a href="http://www.dw.com/en/eu-us-lawmakers-scold-facebook-trump-linked-cambridge-analytica/a-43033278">lawmakers are jumping into the fray</a>, with Zuckerberg expected to testify to Congress. And the scandals don’t stop there: shortly afterward, revelations that Facebook Messenger tracked years’ worth of call and texting data from Android users only added to public concerns and mistrust of the platform.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fgiphy.com%2Fembed%2F1zKdb4WSHgY4QKAsjo%2Ftwitter%2Fiframe&amp;display_name=Giphy&amp;url=https%3A%2F%2Fmedia.giphy.com%2Fmedia%2F1zKdb4WSHgY4QKAsjo%2Fgiphy.gif&amp;image=https%3A%2F%2Fi.giphy.com%2Fmedia%2F1zKdb4WSHgY4QKAsjo%2Fgiphy.gif&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=giphy" width="435" height="406" frameborder="0" scrolling="no"><a href="https://medium.com/media/1bc6dc047ffb9f2d07a365b6d51f3836/href">https://medium.com/media/1bc6dc047ffb9f2d07a365b6d51f3836/href</a></iframe><p>It’s clear in the past few weeks that users are confused and worried — but should botmakers be? It’s easy to compare the data scraped by Cambridge Analytica — which fit users into psychological profiles using a personality quiz app — to the kind of personalization that chatbots depend on to provide a personalized experience.</p><p>So should users be wary around chatbots? What can botmakers do to assuage those fears?</p><p>Data security is incredibly important, especially for botmakers. Thankfully, if your team is taking data security and user privacy seriously throughout your botmaking process (from brainstorming an initial chatbot design to iterating it long after release), neither you nor your users have anything to worry about. Let’s delve into unique challenges in data security that chatbots face, and how you can assure your users in a post-Cambridge Analytica world.</p><h3><strong>Anticipate That Users Will Over-Share</strong></h3><p>It goes without saying that chatbots need data from users to do their jobs. Often, chatbots will ask anything from the innocuous (“What hobbies do you like?”) to the more personal (“What symptoms are you feeling?”). While some chatbots are more personal than others — <a href="http://telegram.me/marfa_bot">Marfa</a>, for example, is tailored as a friend that users can confide in with their biggest hopes and fears — <em>any</em> chatbot is likely to run into a user who unintentionally over-shares sensitive information.</p><p>For example, let’s say you’ve built a chatbot that turns any message into an animated gif. It’s easy to see how this could have the potential to “overhear” or collect sensitive messages. For a more obvious example, consider a self-help chatbot that tracks a user’s mood twice a day, then asks what made them feel a certain way.</p><blockquote>Workplace chatbots that sit in on group threads might not only pick up on personal details but confidential business matters as well.</blockquote><p>When botmakers begin conceptualizing their chatbots and writing out prompts, they must remember that some users will provide more information than necessary. A good way to stop users from oversharing is to make it more efficient in how it asks for information.</p><h3><strong>Consider Who Will See the Data Shared by Users</strong></h3><p>Before your bot goes public, you need to carefully consider who will get access to user data. For example, will you store data locally, or in the cloud (thereby sharing it with a third party)? Similarly, who within your organization is training the chatbot and has access to user conversations? Is conversation being shared to a third-party NLP service?</p><blockquote>When members of your review team have access to conversation data, it can be a good idea to anonymize users.</blockquote><p>This is especially important if conversations are searchable with specific terms. When data is being shared with third parties (and by default it is, because messaging data belongs to the messaging platform as well), ensure you understand what information is collected and how it is used.</p><p>Speaking of platforms, different platforms collect different types of data and use them for different purposes. When in doubt, read over a platform’s <strong>privacy policy</strong> before choosing it. It’s a good idea to include your own privacy policy to the user so they know how you yourself will use the information that’s provided.</p><h3><strong>Protecting User Data is a Great Way to Build Trust</strong></h3><p>While the mishandling of data by Facebook and Cambridge Analytica is unfortunate, this is a great opportunity to discuss privacy with your users. By reminding them that you take their data seriously and explaining what information is collected (and why), you’re building user trust. In doing so, you don’t need to be worried about the recent scandal. Use this as an opportunity to educate your team and your users on data safety, which will help you in the long run.</p><p>Tuba Tezer is the Growth Marketer &amp; Developer Evangelist of <a href="https://medium.com/u/8f2cabaf5f37">Botanalytics</a>, the most comprehensive <a href="https://botanalytics.co/">conversational analytics</a> tool for chatbots and voice interfaces.</p><p>We’re committed to improving human-bot interactions and conversational UI through data. By offering analytics for bots, we empower bot developers, corporate brands and agencies to improve their chatbots with historical and real time conversational data.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f358da3c3e01" width="1" height="1" alt=""><hr><p><a href="https://medium.com/chatbots-journal/cambridge-analytica-has-botmakers-worried-but-should-they-be-f358da3c3e01">Cambridge Analytica Has Botmakers Worried — But Should They Be?</a> was originally published in <a href="https://medium.com/chatbots-journal">Chatbots Journal</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Examining Chatbot Usage by Country Around the World]]></title>
            <link>https://medium.com/chat-bots/examining-chatbot-usage-by-country-around-the-world-e114a2ce8692?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/e114a2ce8692</guid>
            <category><![CDATA[apps]]></category>
            <category><![CDATA[chatbot-usage-stats]]></category>
            <category><![CDATA[chatbots-for-business]]></category>
            <category><![CDATA[chatbot-market]]></category>
            <category><![CDATA[chatbots]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Tue, 13 Mar 2018 09:33:34 GMT</pubDate>
            <atom:updated>2018-05-10T14:24:20.651Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qwksGJe82EyBePaxpxD2ag.png" /></figure><p>Talk about chatbots is almost always focused on large American brands, startups, and enterprises. Large corporations like Google, Amazon and Facebook often dominate the conversation around artificial intelligence, virtual assistants and chatbots. But this focus on American chatbot market analysis leaves out consumers of other regions of the world, each of which have different behaviors and expectations when it comes to chatbots.</p><p>To better understand how chatbots are used around the world, we’re going to explore chatbot usage stats. What does one audience privilege, versus others? How are companies in one country driving growth that might not serve as a key strategy in another region? Before getting into these details, let’s first look at the similarities in chatbot usage across the globe.</p><h3><strong>Global Views on Chatbot Market Growth</strong></h3><p>If there’s one thing the world can agree on, it’s that chatbots are well received everywhere. A global market poll found that <a href="http://www.businessinsider.com/chatbot-market-stats-trends-size-ecosystem-research-2017-10">45% of users</a> around the world preferred chatbots for customer service. In another global chatbot market analysis determined that <a href="http://www.businessinsider.com/chatbots-are-gaining-traction-2017-5">33% of customers</a> had an overall favorable opinion on chatbots; while that number might seem low, it’s significantly higher than the 11% who held negative opinions. 51% of those polled held a neutral view, suggesting lots of potential for growth. Finally, <a href="http://www.businessinsider.com/chatbot-market-stats-trends-size-ecosystem-research-2017-10">80% of businesses</a> polled in France, the UK, South Africa and the Netherlands planned to implement chatbots by 2020, showing a global trend in chatbot adoption.</p><p>That said, some nations are more receptive to chatbots than others. By looking at chatbot usage stats and chatbot market growth around the world, we can determine:</p><ul><li>What users want and expect from bots in a given country</li><li>How companies and industries are driving chatbot market growth to meet those needs</li></ul><p>For example, European consumers are <a href="http://www.businessinsider.com/chatbots-are-gaining-traction-2017-5">more receptive</a> of chatbots than the United States: 50% of French consumers hold a positive opinion on bots, while only 32% of Americans share the view. Americans’ reticence to adopt chatbots might be blamed on their preference for human assistance over bots: 51% of respondents to the same poll preferred humans, the highest preference compared to other countries.</p><p>To get a better sense of chatbot usage by country, we’ll take a chatbot market analysis on key regions in the world, comparing and contrasting chatbot usage and where the most potential for growth lies. What do consumers in one region expect from bots, and how are they using them? How do companies deliver on that expectation?</p><h3><strong>Chatbot Usage in the UK</strong></h3><p>While European consumers seem to love chatbots more than anyone else as mentioned above, the region <a href="https://venturebeat.com/2017/01/20/introducing-the-european-bots-landscape/">lags behind</a> the US in chatbot development. This isn’t necessarily a bad thing; because regional consumers associate positively with bots (and because ecommerce there is so strong), this indicates plenty of room for growth with few giants dominating the space.</p><p>The UK in particular may drive chatbot usage thanks to its speedy adoption of messaging apps. Mobile messaging in the country <a href="https://chatbotsmagazine.com/trends-driving-the-chatbot-growth-77b78145bac">had a far higher adoption rate</a> among businesses in the past four years, outpacing adoption rates for social media and apps. This means UK businesses were quick to understand the potential of reaching consumers via messaging apps; with time to get accustomed to the new medium, they may become the next big arena for chatbot market growth.</p><h3><strong>Chatbot Usage Stats in Asia</strong></h3><p>Asia is where modern chatbot usage was refined; Chinese WeChat bots were <a href="https://chatbotsmagazine.com/china-wechat-and-the-origins-of-chatbots-89c481f15a44">instrumental</a> to integrating chatbots into retail, daily life tasks and social media usage. But how are consumers in other Asian countries using bots each day? With a <a href="https://engineering.linecorp.com/en/blog/detail/203">case study</a> from LINE, the most popular messaging app for the Asian market outside China, we can look at chatbot usage by country.</p><p>In Japan, prompt and polite customer service is important to businesses and consumers. Chatbot usage has developed to reflect that, with many bots providing customer loyalty programs and services that were previously used with cards. In Taiwan, consumers mostly use chatbots for banking and financial services; many major banks in the country are available on LINE making it easy for customers to check their balances and send or receive money. In Indonesia, ecommerce is the favored use case.</p><h3><strong>The Potential Chatbot Market Growth in Africa</strong></h3><p>In many African countries like Nigeria, mobile devices are users’ key to the digital world as opposed to PC’s. Even before businesses around the world scrambled to provide chatbot solutions, many Africans made use of conversational SMS services to exchange money, check market prices and solve everyday problems. M-Pesa is one such service <a href="http://www.cnn.com/2017/02/21/africa/mpesa-10th-anniversary/index.html">that’s existed for over a decade</a>.</p><p>What does this mean? For one, Africans have long been accustomed to using conversational interfaces on mobile devices to accomplish a wide variety of tasks. Chatbot development in the region can help improve quality of life and provide better aid services to underdeveloped nations. For example, the World Food Program is <a href="https://qz.com/1066734/this-aid-agency-is-using-chatbots-to-beat-world-hunger/">using chatbots</a> to gather data on how it can better meet the needs of people on the continent.</p><h3><strong>App Preferences Across the World</strong></h3><p>Botmakers should choose the right platform for their bot based on where their target audience lives. According to <a href="https://chatbotsmagazine.com/trends-driving-the-chatbot-growth-77b78145bac">one study</a>, audiences in the US and the UK favor Facebook Messenger over other messaging apps.</p><blockquote>French audiences, interestingly, prefer SMS, though Facebook Messenger follows closely behind. Germany, India, South Africa, Nigeria and Brazil all prefer WhatsApp. Unsurprisingly, WeChat overwhelmingly dominates the Chinese chatbot market.</blockquote><p>Tuba Tezer is the Growth Marketer &amp; Developer Evangelist of <a href="https://medium.com/u/8f2cabaf5f37">Botanalytics</a>, the most comprehensive conversational analytics tool for chatbots.</p><p>We’re committed to improving human-bot interactions and conversational UI through data. By offering analytics for bots, we empower bot developers, corporate brands and agencies to improve their chatbots with historical and real time conversation data.</p><p>We aim to make it as easy as possible for developers to act on data, and are available to bots spread across 12 different platforms.</p><p>You can <a href="https://botanalytics.co/">sign up for free</a> to enjoy our <a href="https://botanalytics.co/features">features</a>.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F58521e%3Fas_embed%3Dtrue&amp;dntp=1&amp;url=https%3A%2F%2Fupscri.be%2F58521e%2F&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/c90883b0516ddbe5c59960c12a04bc64/href">https://medium.com/media/c90883b0516ddbe5c59960c12a04bc64/href</a></iframe><figure><img alt="" src="https://cdn-images-1.medium.com/max/705/1*mToBIVQSczojw0mSn2h5Sw.gif" /><figcaption>👏👏Clap below to recommend this article to others👏👏</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e114a2ce8692" width="1" height="1" alt=""><hr><p><a href="https://medium.com/chat-bots/examining-chatbot-usage-by-country-around-the-world-e114a2ce8692">Examining Chatbot Usage by Country Around the World</a> was originally published in <a href="https://medium.com/chat-bots">Chatbots Magazine</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Give Your Bot Some Much-Needed Personality]]></title>
            <link>https://medium.com/bottutorials/how-to-give-your-bot-some-much-needed-personality-7946cffa8984?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/7946cffa8984</guid>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[bots]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[conversational-ux]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Fri, 26 Jan 2018 12:34:56 GMT</pubDate>
            <atom:updated>2018-12-07T04:20:05.302Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pMtdy5gfOWei4cBJOlXJbA.jpeg" /></figure><p>Sometimes overlooked, a good personality is essential to a great chatbot user experience. Just like humans, personality quirks are what make a chatbot unique. Furthermore, a chatbot’s personality makes a profound impact on the user and makes that interaction much more memorable, making it an important factor to remember in your <a href="https://botanalytics.co/blog/2017/10/02/develop-chatbot-content-strategy/">chatbot content strategy</a>. Think about the last time you connected with someone on a deep level: what was that like? Delivering a similar experience to users ensures your bot is a pleasure to use. But where to begin with building a chatbot personality, or with determining which personality best fits a bot’s function?</p><h3><strong>The Basics to Personality Building</strong></h3><p>To get started creating your chatbot personality, consider the five factors of personality:</p><ul><li>Extraversion (how assertive vs. reserved your bot is)</li><li>Agreeableness (is the bot cooperative, or quick to disagree?)</li><li>Conscientiousness (is the bot focused on completing a task, or easily distracted?)</li><li>Neuroticism (how anxious your bot is)</li><li>Openness (whether your bot is open to new ideas or narrow-minded)</li></ul><p>These factors are essential to determining the basic personality map of the chatbot. When considering these factors, keep your target audience and the bot’s goal in mind. Who is likely to use the bot, and why? When? A bot helping younger users order food late at night will have a different personality than one that provides the morning news to business professionals.</p><p>Try writing different scripts to test voices, tones and attitudes that will make up the bot’s persona. A great way to do this is to imagine you’re making a pitch directly to the target audience. What kind of language and diction would you use to appeal to them? Write down everything you can think of to get a head-start on developing your chatbot personality. You might even role-play with a member of the target audience by chatting directly with them in that voice to see how it works in-action. Once your brainstorm is complete, share your scripts with other members of the team. They’ll help you polish the chatbot’s persona and suggest changes where necessary.</p><h3><strong>Conversational Tips for More Personality</strong></h3><h4><strong>Make Small Talk</strong></h4><p>If your chatbot isn’t focused on accomplishing a task as quickly as possible, inject some small talk and opportunities to show off personality quirks. Anticipate banter like “Tell me a story,” “What can you do?” and “What’s up?” These queries can be difficult to account for, but as you iterate your bot, you can add responses as users ask for them. These off-topic queries help showcase your bot’s personality and make it feel like a more fleshed out character.</p><p>You may also want to keep the chatbot up-to-date with seasonal or current events. This means greeting users appropriately for holidays or giving it some banter related to a big event — like the latest music album drop that <em>everyone’s</em> talking about. By embellishing your responses with these comments, the chatbot will have its finger on the pulse of society. Accomplishing this well, though, means really knowing what’s important to your audience and what they’re likely to talk about.</p><h4><strong>Talk the Talk</strong></h4><p>A bot feels much more natural and personable if it speaks its users’ language. This means understanding the latest slang and abbreviations that its audience are likely to use. The bot itself should add these terms to its vocabulary and use them as well, providing a more casual experience akin to talking to a close friend who “just gets it.” Account for typos as well to keep the bot from breaking user immersion.</p><p>If the chatbot does run into errors, is should simply ask for clarification in a polite way rather than send the user through annoying conversation loops. Ease the frustration of running to an error by giving the bot a self-deprecating attitude when it messes up.</p><h4><strong>Learn Over the Course of Conversation</strong></h4><p>When we shop often at local businesses, we often form relationships with the people who work there. This means the local bartender or employee at the deli counter already knows what we’re likely to order without having to say a thing. They might even make recommendations based on our preferences when we want to try something new. Of course, chatbots are great at providing this kind of service.</p><p>Your chatbot must actively get to know the user on a deep level, and make the user aware whenever possible that they recognize them and have built a rapport. This not only makes it feel more personable, but can help to increase engagement as the user forges a relationship with the character.</p><h4><strong>Establish Conversation Options</strong></h4><p>Using closed questions rather than open-ended ones is a famous tenet of conversational design. This not only keeps the conversation on-track with fewer errors, but offers a way to showcase your chatbot’s personality as well. For example, button options can be written in a tone/style of voice which matches that of the bot, giving bot and users a shared vocabulary that sets the tone of conversation.</p><p>Conversation options also help the bot steer the conversation down certain paths. How controlling is your bot, or how likely is it to diverge onto tangents at the users’ whim? Consider the factors of agreeableness and conscientiousness as detailed above to help you determine how much control your bot will give the user with closed versus more open responses.</p><p><a href="https://www.linkedin.com/in/tubatezer/">Tuba Tezer</a> is the Growth Marketer &amp; Developer Evangelist of <a href="https://medium.com/u/8f2cabaf5f37">Botanalytics</a>, the world’s leading conversational analytics tool for conversational interfaces.</p><p>We’re committed to improving human-bot interactions and conversational UI through data. By offering analytics for bots, we empower bot developers, corporate brands and agencies to improve their chatbots with historical and real time conversation data.</p><p>We aim to make it as easy as possible for developers to act on data, and are available to bots spread across 13 different platforms.</p><p>You can <a href="https://botanalytics.co/">sign up for free</a> to enjoy our <a href="https://botanalytics.co/features">features</a>.</p><p><a href="https://northstack.com/?utm_source=page&amp;utm_medium=banner&amp;utm_campaign=botslinks">NorthStack™ Serverless Application Hosting</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7946cffa8984" width="1" height="1" alt=""><hr><p><a href="https://medium.com/bottutorials/how-to-give-your-bot-some-much-needed-personality-7946cffa8984">How to Give Your Bot Some Much-Needed Personality</a> was originally published in <a href="https://medium.com/bottutorials">Dev Stash</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Designing for Conversational UI: Humanizing Chatbots]]></title>
            <link>https://medium.com/chatbots-journal/designing-for-conversational-ui-humanizing-chatbots-f199e59b363e?source=rss-8487a91a2673------2</link>
            <guid isPermaLink="false">https://medium.com/p/f199e59b363e</guid>
            <category><![CDATA[bots]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[conversational-design]]></category>
            <category><![CDATA[conversational-ui]]></category>
            <dc:creator><![CDATA[Tuba Tezer]]></dc:creator>
            <pubDate>Thu, 21 Dec 2017 09:16:15 GMT</pubDate>
            <atom:updated>2017-12-21T14:50:19.039Z</atom:updated>
            <content:encoded><![CDATA[<p>With advancements in AI and machine learning, we’re closer to humanlike digital assistants than ever before. As seen with the Turing test, the hope for a chatbot indistinguishable from a human has been around since artificial intelligence’s early days. The reason why humanlike bots are so desired is because they provide a truly seamless, natural and intuitive way to interact with machines; the most sophisticated humanlike chatbots (like Alexa, Google Assistant and Siri) can even pull together tasks from a variety of apps into just one interface.</p><h4><strong>Is Humanizing Chatbots Necessary?</strong></h4><p>Humanizing chatbots make it feel as if a bot is <em>actually listening</em> and responding to us. Lack of intonation, emotion and understanding, meanwhile, can make it feel like a bot is just reading or relaying information, leading to impersonal exchanges. A humanlike chatbot can also make data collection and personalization feel more natural (the difference between getting to know a new friend versus being surveilled by a corporation). Finally, empathetic and natural-sounding bots can take advantage of “<a href="https://www.thinkwithgoogle.com/marketing-resources/micro-moments/micro-moments-understand-new-consumer-behavior/">micro moments</a>” to cross or upsell according to user need.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-xr_asdzIqTtzvyFhh3ajQ.jpeg" /><figcaption>Lovely Wall-E</figcaption></figure><p>That said, humanizing chatbots can have a couple downsides. The most major is that keeping things too open-ended can be confusing to users and cause anxiety, causing a Paradox of Choice. Bots that are more utilitarian and less chatty can also surface up results to users faster. Not everyone has time to navigate small talk with a chatbot! Thankfully, this is a minor con; if your humanlike chatbot is engaging enough, it shouldn’t be an issue.</p><h3><strong>BASIC CONVERSATIONAL UI DESIGN FOR HUMANIZING CHATBOTS</strong></h3><h4><strong>Understand the Rules of Conversation for Humanlike Chatbots</strong></h4><p>You might not realize it, but conversations follow an implicit set of rules. We often take these rules for granted; for the most part, we pick up on subtle social cues without thinking. But if you’ve ever neurotically scrutinized a conversation — for example, <em>should</em> you send another message to your crush even if they hadn’t responded to the last one? — then you’re acutely aware of the basic rules and building blocks of conversation. These are <strong>constitutive</strong> rules (for interpretation and sense-making) and <strong>normative</strong> rules (understanding the appropriate way to respond) — both of which a humanlike chatbot must use in their conversational UI design.</p><p>The first unit of conversation is <strong>turn-taking</strong>. Conversation partners take turns in a conversation: Person A says something, then Person B responds, Person A answers back, etc. The opposite of this is rude behavior: interrupting and talking over each other.</p><p>The next is <strong>threading</strong>. This is the conversation “thread,” or the arc from one topic of conversation to another. As a conversation moves forward, each step contextually links to the one before and after. A lack of threading leads to confusing, nonsensical messages and non-sequiturs. A chatbot’s conversation thread helps the botmaker — and forgetful or confused user — understand the conversational path and flow. This can further help in developing and iterating conversational UI design.</p><p>Responding appropriately to contextual information brings us to the next unit of conversation — and one that the most robotic, on-rails chatbots consistently fail at. A humanlike bot should build on what the user said, then prompt them to move the conversation forward. It should use <strong>syntax</strong> to <strong>intuitively respond to what’s said</strong>. We often adopt a shorthand way of speaking rather than make full, complete statements. Conversation is full of verbal shortcuts, and we’re able to understand what others say due to context.</p><blockquote><strong><em>User:</em></strong><em> What does my day look like?</em></blockquote><blockquote><strong><em>Bot: </em></strong><em>You have a meeting at Bryant Park today at 3pm.</em></blockquote><blockquote><strong><em>User:</em></strong><em> Should I bring an umbrella?</em></blockquote><blockquote><strong><em>Bot:</em></strong><em> It looks like it’s going to rain today in New York.</em></blockquote><blockquote><strong><em>User: </em></strong><em>Is there a café nearby?</em></blockquote><blockquote><strong><em>Bot:</em></strong><em> There is Blue Bottle Coffee on 54 West 40th Street.</em></blockquote><blockquote><strong><em>User: </em></strong><em>Message Greg and tell him to meet me at Blue Bottle Coffee instead of Bryant Park.</em></blockquote><blockquote><strong><em>Bot:</em></strong><em> Got it.</em></blockquote><p>Through contextual cues, the bot can understand what the user is asking without them giving all the details: regardless of the user’s current location, they’re asking for weather and a café near the location of the meeting. Getting this information quickly and seamlessly, the user can then schedule a change in location for the event to keep dry.</p><blockquote>Humanlike chatbots <strong>pay close attention to conversational threads and use natural language processing to tease out the idea behind statements and fragments that would be confusing on their own</strong>.</blockquote><p>This becomes more important with voice-activated and verbal conversational UI design, because we often make fragmentary responses when speaking verbally.</p><p>Strategies using syntax include word substitutions, a chatbot can answer multiple ways to the same query. For example, saying “hi” to a bot might return “Hello,” “Hey there!” or “What’s up?” Answering in response to time-based messages is even more sophisticated. A bot might respond to a user who says, “Good night” with a sleeping emoji, for example.</p><p>Another important aspect to contextualize conversation has to do with error responses. Saying something like “Sorry, I don’t understand” doesn’t help the user understand how to better reword their query. To anticipate the <strong>variable path</strong> of conversation — or when the user acts in an unexpected way — a humanlike chatbot should <a href="https://developers.google.com/actions/design/conversation-repair">ask clarifying questions, rephrase queries and provide new prompts for solving errors.</a></p><h3><strong>STRATEGIES FOR HUMANIZING CHATBOTS</strong></h3><p>In addition to teaching your bot to understand how humans talk with the information above, there are numerous small strategies you can use for humanizing chatbots. The first is to give it emotional understanding. Sentiment analysis enables an AI to <strong>understand the emotion</strong> behind a message, and how strongly that emotion is felt. It can then respond appropriately depending on how the user feels.</p><p><a href="https://woebot.io/">Woebot</a> is a humanlike chatbot that provides therapy via Facebook Messenger, and pulls off emotional understanding with panache: it checks in daily with how a user is feeling, then responds to that feeling in an emotionally empathetic way. It uses <strong>non-verbal responses</strong> like emojis and gifs to further convey emotion. When a user responds correctly to a quiz about mental health, for example, Woebot might respond with a gif of a robot dancing. All of this is incredibly helpful to the bot, because its whole point is to help the user engage with and understand their emotions.</p><p>Another strategy for more humanlike chatbots is to use <strong>filler words</strong>, or non-conversational sounds. These are the “uhhs,” “ums” and “hmmmms” we often pepper conversations with. While it might sound contrived to design a chatbot to respond this way, it does help conversations sound more natural (especially for verbal bots that have a voice). Poncho uses these utterances to make humorous asides. When given a user location to make a weather forecast, it might respond: “Oh yeah? My ex is from there. Hm…anyway…it’s 55°F and partly cloudy there rn.”</p><p>You’ll also notice <a href="https://www.messenger.com/t/hiponcho">Poncho</a> doing something else in that line. It’s using <strong>colloquial slang</strong>, replacing “right now” with internet shorthand, “rn.” Allowing a bot to reflect the way people chat with each other online makes it more relatable and eases confusion when the user speaks in slang.</p><p>Finally, while it doesn’t apply to all bots, <strong>facial expressions</strong> further make for a more humanlike chatbot. Mobile chat platforms don’t allow for dynamic avatars, but if developing an interface of your own, consider having a model <a href="https://www.twitch.tv/mitsuku_irl">like Mitsuku</a>. Facial expressions and body language is a lifelike way to convey non-verbal responses and help the user empathize with the bot.</p><p>Tuba Tezer is the Growth Marketer &amp; Developer Evangelist of <a href="https://medium.com/u/8f2cabaf5f37">Botanalytics</a>, the world’s leading conversational analytics tool for chatbots.</p><p>We’re committed to improving human-bot interactions and conversational UI through data. By offering analytics for bots, we empower bot developers, corporate brands and agencies to improve their chatbots with historical and real time conversation data.</p><p>We aim to make it as easy as possible for developers to act on data, and are available to bots spread across 12 different platforms.</p><p>You can <a href="https://botanalytics.co/">sign up for free</a> to enjoy our <a href="https://botanalytics.co/features">features</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f199e59b363e" width="1" height="1" alt=""><hr><p><a href="https://medium.com/chatbots-journal/designing-for-conversational-ui-humanizing-chatbots-f199e59b363e">Designing for Conversational UI: Humanizing Chatbots</a> was originally published in <a href="https://medium.com/chatbots-journal">Chatbots Journal</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>