<?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 Chang Xu on Medium]]></title>
        <description><![CDATA[Stories by Chang Xu on Medium]]></description>
        <link>https://medium.com/@changxu?source=rss-84ada277731c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Ur-ul0lsb9XMSyX_RVzgKw.png</url>
            <title>Stories by Chang Xu on Medium</title>
            <link>https://medium.com/@changxu?source=rss-84ada277731c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 13:53:56 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@changxu/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[Bad Ideas are Good Again]]></title>
            <link>https://medium.com/basis-set-ventures/bad-ideas-are-good-again-bd9f348f9067?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/bd9f348f9067</guid>
            <category><![CDATA[generative-ai]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[entertainment]]></category>
            <category><![CDATA[gaming]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Wed, 03 May 2023 14:31:34 GMT</pubDate>
            <atom:updated>2023-05-03T14:31:34.160Z</atom:updated>
            <content:encoded><![CDATA[<p>Now that we have large language models (LLMs) and generative AI, as an AI fund, we get asked all the time “what do you invest in now?”</p><p>One answer: “Bad Ideas.” No seriously.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/512/0*CnbmcN3xOQgg6DnG.png" /><figcaption>Because walking off a cliff is probably a bad idea… (Art generated by DALL-E)</figcaption></figure><p>Venture has become almost boring. Many categories are played out, crowded, and seeing diminishing returns. It is difficult for a new idea to be truly fresh and to break out. However, generative AI is resetting everything, across the board. All this new technology is forcing us to challenge fundamental assumptions around how industries operate. As a result, many of what we used to consider Bad Ideas are now Good again. We are excited for what’s to come and to invest in seeds today that will shake up entire industries. There are a number of ways this is happening. Let’s dive in.</p><h3>Bad categories are now Good</h3><p>Gaming, entertainment, and media have been considered dead categories for venture investments because of their return profile. Gaming studios and media IP franchises are well-known as hits-driven businesses, which makes it difficult to either invest directly or into businesses that sell into them. Generative AI fundamentally disrupts the cost structure of these businesses, which will shake up the industry and create opportunities for big winners.</p><h3>The disruption</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*GgXRIxzXpi-Mobr3.jpg" /><figcaption>Motion capture is an example of a laborious and expensive technique used by gaming and entertainment studios to achieve lifelike animations for 3D characters</figcaption></figure><p>The major cost centers in these businesses — where the most time, money, and resources are spent today — are precisely where generative AI is best applied to. For a gaming studio, it is the labor-intensive process of creating art, 3D models and animation, which grows more content to monetize. For Hollywood, it is visual effects (or VFX), using many techniques to bring to life on screen what’s not present in real life, whether it be fantastical characters, scenes or motion.</p><p>AI has conquered 2D imagery (with Stable Diffusion, Midjourney, and DALL-E etc), and the frontier is moving towards 3D models and animation. Today, an animated feature needs 24 frames per second and each frame needs to be created by hand. Tomorrow, such works can be generated in minutes with AI. Today, to make a 3D character look and move with realistic human motion requires the manual workflows of rigging, motion capture, and frame-by-frame cleanup. Tomorrow, AI can generate realistic motion in real-time that abides by the laws of physics and is tailored to each character.</p><p>A number of tech players are working on the many aspects of this and selling into the incumbents. To name a few, <a href="https://zivadynamics.com/">Ziva</a> focuses on realistic facial motion, <a href="https://wonderdynamics.com/">Wonder Dynamics</a> focuses on CGI for Hollywood, <a href="https://uthana.com/">Uthana</a> focuses on animation in games, <a href="https://www.mirageml.com/">Mirage</a> focuses on generating 3D scenes, and <a href="https://volinga.ai/">Volinga</a> applies neural radiance fields (or NeRFs) to virtual production. All will drastically reduce the cost of production or make the impossible, possible.</p><h3>Rise of the indies</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lS3gaQypoADCmKz69UjP2w.png" /><figcaption>Roblox has become so successful because it empowers anyone to create games</figcaption></figure><p>Disruption will also come from below and we’re seeing this in the rise of the indies. When the cost to produce asymptotes to zero, every kid in their bedroom can create games and films; the limit is only their imagination. These indie creators will be more forgiving of imperfections as the tech improves and they’ll be more willing to try new tools vs professionals in industry. New form factors for entertainment will rise, not limited by today’s constraints in production. The successes of Roblox, Minecraft, and Fortnite have shown us what is possible when you make a game that enables players to create games.</p><p>Many startups are already working on this as well. For example, <a href="https://lumalabs.ai/">Luma AI</a> applies NeRFs to empower anyone to create photorealistic game assets by just capturing a cell phone video. <a href="https://versed.game/">Versed</a>, <a href="https://dreamlab.gg/">Dreamlab</a>, and others allow anyone to make games by generating scripts, assets, levels, and animation. Versed and <a href="https://aidungeon.io/">AI Dungeon</a> are reimagining role-playing games, starting with the fanatic Dungeons and Dragons community. <a href="https://www.linum.ai/">Linum</a> empowers anyone to generate video. <a href="https://www.talofagames.com/">Talofa</a> is building the Roblox of fitness gaming for gen Z. We will see an explosion of creation.</p><h3>Bottleneck shifts will create new winners</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*MqmPwSGHI1hrjRgY" /><figcaption>Video game play testers (in the Apple TV+ series Mythic Quest)</figcaption></figure><p>In a future where labor for tasks like animation, VFX and 3D modeling are no longer bottlenecks, the plane of competition will shift. When a AAA studio can make 100 high quality games per year vs one and when indies can create high quality games and films without big budgets, the bottleneck will no longer be studio capacity, but rather the capacity of humans to create and consume higher-quality content and new and more personalized experiences.</p><p>In 5–10 years, the leaderboards for AAA studios and top Hollywood studios will be reshuffled in a major way. The ones today that cannot or will not adapt to this oncoming revolution will fade. Tech is advancing rapidly on many fronts and while it’s premature to predict exactly where the value will accrue, we can take a hint from history.</p><p>No industry is perhaps more acquainted with the implications of technological disruption than the entertainment industry. From the advent of the talkies to streaming, every new technological innovation has created new winners and losers, new dimensions of competition and ultimately stronger content and more optionality for consumers. Today’s shifts will propel new layers in the stack and new platforms to prominence, creating new winners. Technology won’t alleviate work, it will merely elevate expectations and incentivize competition in new ways.</p><h3>Call for Bad Ideas</h3><p>Generative AI is changing the game for venture investments, enabling us to look at categories previously full of Bad Ideas with fresh eyes. Gaming, entertainment, media, and many other areas that were once seen as dead categories are now becoming attractive.</p><p>We are at a unique point in time where the laws of physics suddenly work differently. The assumed advantages for incumbent players are going away and the rules of how industries have worked can be challenged. By fundamentally disrupting the cost structure of these businesses, generative AI is creating opportunities for big winners and shaking up entire industries. The rise of indies, the creation of unique form factors for entertainment, and the emergence of new bottlenecks will all create fresh winners, leading to a major industry reshuffling in the coming years.</p><p>As an early stage VC, we feel so much excitement for what’s to come. We will continue to share our thoughts in the coming weeks. If you’re working on a Bad Idea (that you believe now to be Good Idea), we would love to hear it! Shoot me a note at chang@basisset.com</p><p><em>Many thanks to Matt Aimonetti and Eduardo Fonseca for reading drafts of this post and providing valuable feedback.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bd9f348f9067" width="1" height="1" alt=""><hr><p><a href="https://medium.com/basis-set-ventures/bad-ideas-are-good-again-bd9f348f9067">Bad Ideas are Good Again</a> was originally published in <a href="https://medium.com/basis-set-ventures">Basis Set</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[All That Glitters (Like Stars) Is Not Gold: Metrics that matter when commercializing open source]]></title>
            <link>https://medium.com/basis-set-ventures/all-that-glitters-like-stars-is-not-gold-metrics-that-matter-when-commercializing-open-source-2f58c8a5b493?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f58c8a5b493</guid>
            <category><![CDATA[vc]]></category>
            <category><![CDATA[dev]]></category>
            <category><![CDATA[github]]></category>
            <category><![CDATA[open-source]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Fri, 11 Dec 2020 14:50:05 GMT</pubDate>
            <atom:updated>2020-12-11T14:50:05.591Z</atom:updated>
            <content:encoded><![CDATA[<p>The monetization of open source is laying the foundation for the next set of iconic tech companies. As open source has matured from an ideology into a go-to-market motion, category-defining companies are being built on open source, like MongoDB, Elastic and GitHub, to name a few.</p><p>Since late 2017 when we led the seed round for <a href="http://rasa.com/">Rasa</a>, an open source conversational AI platform, we’ve been tracking and working to understand the most important drivers behind open source successes. We analyzed data from the hottest new open source companies to help founders make sense of early metrics and determine what to prioritize. We are starting with data on GitHub, where almost all open source projects live, and will layer on data from other channels — e.g., downloads, installs, Hackernews, Reddit, Stackoverflow, Twitter — in future articles.</p><p>For founders building open source companies, they must continuously and accurately assess the awareness and engagement of external developers, who adopt your product as users and some may also become contributors. Adoption is universally important. Getting contributions from your community comes into play later on and may be less important depending on your product, sector and strategy. Nothing is more telling of a company’s ability to win on these dimensions than <strong>Issues</strong>, <strong>Serious Contributors</strong>, or the <strong>Contributor Coverage Ratio</strong>.</p><h3>Open source: moving from exception to mainstream</h3><p>The <a href="https://airtable.com/shrOYJoe8AZn1A5eU"><strong>Basis Set Open Source Index</strong></a> includes 134 companies founded since 2015 that are commercializing open source and have raised at least $1M in venture funding. More open source companies are founded every year, from 21 in 2015 to 26 in 2019 [1].</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aHQsjC_64yx322uyIX7ecw.png" /><figcaption>See the full list of companies in the <a href="https://airtable.com/shrOYJoe8AZn1A5eU">Basis Set Open Source Index</a></figcaption></figure><p>More companies are also deliberately using open source as go-to-market strategy. In 2015, 67% of companies that launch open source launch it much later after company formation. By 2020, this statistic has drastically dropped to 15%. By contrast, the proportion of companies that were formed concurrently as their open source project have risen from 5% to 38% [2]. Companies are often reaching for open source right out of the gate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/946/1*51Qm1dL4gfkPNrCfs2uEaA.png" /><figcaption>More companies are using open source from day one</figcaption></figure><p>As a rough proxy for success, we used total funding raised [3]. We looked at how companies’ open source repos perform across different funding tiers from $1–5M raised to $50M+ raised, roughly correlating with seed stage to growth stage.</p><h3>I got 99 Issues and that’s a good thing</h3><p>When looking at adoption, <strong>Issues</strong> is the name of the game. It grows the fastest and most consistently among all the metrics, because it shows that external developers are actually engaging with your product. Growth stage companies typically have 5x the number of <strong>Issues</strong> as seed stage companies.</p><p>Rick Lamers, the cofounder &amp; CEO of <a href="http://orchest.io/">Orchest</a>, mentioned that when he was building the popular open source project <a href="https://github.com/ricklamers/gridstudio">Grid Studio</a>, he noticed that the number of <strong>Issues</strong> jumped at around 6,000 stars. It was clear that he had active users and they were running into real issues.</p><p>As you build in public, you can naturally have your <strong>Issues</strong> flow into your product roadmap publicly, as <a href="https://github.com/github/roadmap/projects/1">GitHub</a> and <a href="https://github.com/cockroachdb/cockroach/projects">Cockroach</a> do. Other best practices for engaging your community include writing high quality documentation, detailed release notes, like <a href="https://gist.github.com/spinnaker-release/94280a2b615adccd975eed73359023ac">Spinnaker</a>, or changelogs, like <a href="https://rasa.com/docs/rasa/changelog/">Rasa</a> and <a href="https://github.com/pulumi/pulumi/blob/master/CHANGELOG.md">Pulumi</a>. Being open with your community fosters trust.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CId_Xd7ts493u-KdlxTaVQ.png" /><figcaption>To understand adoption, Issues is the most important metric</figcaption></figure><p><strong>Forks</strong> grow the next fastest, at 4x, because it indicates a bit more active engagement than stars or watchers. It is a prerequisite for actually playing with the code or building something on top of it.</p><p><strong>Stars</strong> and <strong>Watchers</strong> grow the most slowly and have more variability, at 2–3x. They’re perhaps the least interesting as metrics because they track interest passively.</p><p>In conversation with <a href="https://twitter.com/jasoncwarner">Jason Warner</a>, CTO of GitHub, he offered a fun analogy for understanding this: there are two inputs to a computer — a keyboard and a mouse. He values the keyboard and doesn’t value the mouse. Typing up <strong>Issues</strong> indicates a far stronger interest than clicking to <strong>Star</strong> a repo.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1018/0*uWfuzLGC_gEuHXBU.png" /><figcaption>The keyboard (<strong>Issues</strong>) is more valuable than the mouse (<strong>Stars</strong>)</figcaption></figure><h3>Contributors and the Coverage Ratio: the inside scoop</h3><p>Moving from adoption to contribution often signals the next level of engagement, though some projects strategically choose to focus only on user growth and not on contributor growth, such as Babel [4]. In looking at metrics about contribution, we excluded the companies that were open source-first as their original open source communities — some very large and they often do not have a lot of control over — gave them very different starting points.</p><p>The number of <strong>Contributors</strong> grows at the fastest rate. In fact, the number of <strong>Serious Contributors</strong> (who have made &gt;2 commits) rises sharply as companies get into the growth stage. Growth stage companies have 9x as many <strong>Contributors</strong> and 8x as many <strong>Serious Contributors</strong> as seed stage companies. Pull Requests and Commits grow at a more modest rate of 3–4x.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-eSwknBl99f0sBHw1nUxcA.png" /><figcaption>To understand contributors, Contributor and Serious Contributor count are the most important metrics</figcaption></figure><p>To understand why contributor growth is so important, we looked at the <strong>Contributor Coverage Ratio</strong>, that is, the ratio of contributors to employees, for companies that are open source-first vs. those that are company-first or started at the same time. At seed stage, those that are open source-first had a significant advantage where they had a 23x <strong>Contributor Coverage Ratio</strong>. But by growth stage, this advantage has evaporated as both groups have <strong>Contributor Coverage Ratios</strong> just over 2–3x. If building a community of contributors is important to the strategy, then to be successful, companies that leverage open source as a go-to-market strategy need to be able to attract contributors at a rate commensurate with their company growth. Even those that are open source first can’t simply focus on commercialization and need to continue to invest in community.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PSoXVMaUSPRMojrAorY6eQ.png" /><figcaption>Contributor growth should keep up with company growth</figcaption></figure><p>Successful open source companies view their community as a source of leads for customers, talent, and support. Rasa recognizes that it’s crucial to <a href="https://rasa.com/community/contribute/">attract contributors</a>, so they make it easy to contribute and celebrate contributors, leading to an excellent <strong>Contributor Coverage Ratio</strong> of 4x. In some cases, we’ve heard about companies strategically leaving unfinished features in their code so that avid users would be enticed to contribute! On the other hand, if the open source community is not growing, then that would also starve the growth of the commercial side and lead to a painful reckoning.</p><h3>Three metrics to rule them all</h3><p>Open source is powerful because fostering a developer community drives both development velocity as well as bottoms-up adoption for enterprise sales, thereby creating a formidable moat and often enabling these companies to scale faster than their closed competitors. But open source complicates the business model by adding more degrees of freedom as well as inviting more public scrutiny on how you build your company. It is very powerful if you get the magic formula working.</p><p>As food for thought, here is a framework for early stage founders building open source companies to think about what GitHub metrics to focus on. This is an alpha version, we are continuing to iterate and would love any feedback.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KJK16us8oIWUGc5oh9u9WA.png" /><figcaption>Basis Set Framework for Open Source Metrics on GitHub</figcaption></figure><p>Stay tuned for future articles in this series where we layer on data from other channels.</p><p>We at <a href="http://basisset.ventures/">Basis Set Ventures</a> are excited about building the next generation of category leaders via open source. If you have an early stage company or thinking of forming one from your open source project, shoot us a note at <a href="/service/mailto:chang@basisset.ventures">chang@basisset.ventures</a>.</p><p><em>Thank you to </em><a href="https://twitter.com/mattetti"><em>Matt Aimonetti</em></a><em> for reading a draft of this article and </em><a href="https://twitter.com/jasoncwarner"><em>Jason Warner</em></a><em> for providing valuable feedback.</em></p><h3>Methodology</h3><p><a href="https://airtable.com/shrOYJoe8AZn1A5eU">Sample</a>:</p><ul><li>~134 companies founded on or after 2015 that have raised at least $1M in total funding and are still privately held. See the list of companies included in the <a href="https://airtable.com/shrOYJoe8AZn1A5eU">Basis Set Open Source Index here</a>. If we’ve missed your company, <a href="https://airtable.com/shrk8wfjlDZ0nKxul">let us know here</a>.</li><li>…with an open source component to their business. We included both companies for whom open source is core to their strategy (e.g., Fishtown Analytics) as well as those for whom open source is ancillary (e.g., Replicated), but not trivial.</li></ul><p>Linking company to GitHub repo:</p><ul><li>As companies frequently have many repos on GitHub, we looked at the repo with the most number of stars and designated it as the primary repo for each company</li><li>We designated GitHub repo by project origin, even if the OSS is not owned by the company, eg Astronomer/Airflow, StreamNative/Pulsar, Altinity/Clickhouse, Ahana/Presto, and Starburst/Presto</li><li>Where applicable, if a company arose out of an open source project and then created a new open source project, we chose to look at the original open source project, eg Observable/D3</li></ul><p>Data sources:</p><ul><li>Crunchbase, GitHub, Pitchbook, company websites, LinkedIn, company meetings, manual cleaning</li></ul><h3>Footnotes</h3><p>[1] Some 2020 companies are likely still in stealth or haven’t yet raised $1M+, so the count will increase</p><p>[2] “Same time” — defined as company formation and open source founding (first commit to GitHub repo) is within 3 months of each other.</p><p>[3] We understand that funding is an imperfect metric and there are many nuances, but it is objective and correlates with the viability and success of the business.</p><p>[4] Nadia Eghbal calls these projects with high user growth and low contributor growth “Stadiums” in her book “Working in Public: The Making and Maintenance of Open Source Software.”</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f58c8a5b493" width="1" height="1" alt=""><hr><p><a href="https://medium.com/basis-set-ventures/all-that-glitters-like-stars-is-not-gold-metrics-that-matter-when-commercializing-open-source-2f58c8a5b493">All That Glitters (Like Stars) Is Not Gold: Metrics that matter when commercializing open source</a> was originally published in <a href="https://medium.com/basis-set-ventures">Basis Set</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Working but Not at Work: The Future Workplace and Nexkey]]></title>
            <link>https://medium.com/upfront-insights/working-but-not-at-work-the-future-workplace-and-nexkey-5fdb83daecc3?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/5fdb83daecc3</guid>
            <category><![CDATA[coworking]]></category>
            <category><![CDATA[portfolio]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Wed, 26 Sep 2018 16:10:52 GMT</pubDate>
            <atom:updated>2018-09-26T16:10:52.701Z</atom:updated>
            <content:encoded><![CDATA[<p>On weekdays, my coworkers scour the neighborhood for new coffee spots for the 3pm pick-me-up. This time, they found a cool one: <a href="https://www.rapha.cc/us/en_US/clubhouses/los-angeles">Rapha</a> is a bike shop that is also a coffee shop. Surrounded by walls of biking gear, the center island in the store sells coffee from an expensive-looking espresso machine and pastries, next to it a communal table with chairs as well as a couple smaller tables for a more intimate setting. It’s not a shop, it’s a “Clubhouse,” the barista-cashier quips.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*JWeCMK4pj18tuuh1cgpoJg.jpeg" /><figcaption>Rapha in downtown Santa Monica is a “Clubhouse” for cyclists that also sells coffee and offers a workspace. Image credit: Menemsha.</figcaption></figure><p>Parting with $200 for an annual membership will get you free coffee for a year, which is incentive enough for my partner <a href="https://twitter.com/kevinyzhang">Kevin</a>, who now strives to hold all his coffee meetings there. <a href="https://twitter.com/kftaylor">Kyle</a>, who’s actually a biking enthusiast, uses it for its intended purpose — being able to rent high-end bikes cheaply. It’s a cool place to hang out and work for a few hours if you’re into biking or if you’re just a young professional.</p><p>Rapha is a trailblazer of a growing trend. Physical spaces are rethinking what they do and what they offer. Many shops today are still trying to turn over customers as quickly as possible by not offering ample seating, WiFi, wall outlets for charging, or even bathrooms. Others are recognizing the opportunity: tired of traditional offices, people want more flexibility and variety in the spaces in which they work and socialize.</p><p>After all, what activity do we spend more time doing during the week than work?</p><h3><strong>The coworking explosion</strong></h3><p>For proof of this trend, look no further than coworking, which has been exploding in the last few years. <a href="https://www.dropbox.com/s/rjbmdo4wp4aeccx/2018%20Complete%20Coworking%20Forecast.pdf?dl=0">Deskmag</a> projects ~19,000 coworking spaces supporting more than 1.7 million numbers by the end of 2018, which is 2–3 times of the ~9,000 coworking spaces and 545,000 members in just 2015. WeWork was only founded in 2010 and it is already Manhattan’s largest tenant, managing <a href="https://www.6sqft.com/wework-is-officially-manhattans-largest-private-office-tenant/">5.3 million square feet of office space</a>.</p><p>You probably think that coworking is all about the gig economy and early-stage startups, that millennials are renting desks or cozy private offices because they are lured by the foosball tables and beer on tap. But it is about a lot more than that.</p><p>It is about corporates wanting the flexibility of having both long-term real estate as well as a short-term low-barrier-to-entry workspace that they do not even need to manage. It is about shifting from CapEx to OpEx and being able to scale up or down easily, as well as cutting costs as office rents continue to rise. Coworking enables corporates to have a real estate strategy that can adapt quickly as their business needs change.</p><p>It is also about millennial workers who prefer workspaces that have better amenities and services and a more relaxed environment. They want the flexibility of going to work at more convenient locations than the usual corporate campuses in the suburbs, as they tend to live in cities and loathe long commutes. Working remotely could be terribly lonely and they prefer the social interactions and community aspects of coworking.</p><h3><strong>The rise of multifunctional spaces</strong></h3><p>Traditional storefronts are catching on and it is not just Rapha.</p><p>Down the street, <a href="https://www.businessinsider.com/inside-capital-one-cafe-for-millennials-2017-2">Capital One opened a cafe</a> in order to appeal to millennials in a different way. It offers Peet’s Coffee, free WiFi, device charging stations, and comfy seating to encourage extended stays. It is a radical departure from the typical bank parlors, with no-nonsense security guards flanking the entrance, where people line up to cash checks or get squirreled away to a cubicle for more complicated matters. The only signs of the bank that is operating this space are product promotions that bedeck the walls as well as a room where you can get free coaching on your money problems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*fvXpHogSQ3wpE-iCgkJGhg.jpeg" /><figcaption>Capital One Cafe reimagines the traditional bank parlor as a coffee shop and workspace. Image credit: Studio J Marketing.</figcaption></figure><p>But, none is more creative than Metro Cafe, which is a stone’s throw away from our office. It is a coffee shop, <a href="http://metrocolab.org/">a coworking space</a>, and… a <a href="https://togetherla.net/metro-calvary-chapel-metropolis-beyond-sunday/">church</a>! The pastor was inspired by a vision that his church should bring people together over coffee and foster collaboration, quite literally. To make it even more unusual, there are no listed prices and you pay what you want for the coffee, including nothing.</p><p>While Downtown Santa Monica is demonstrably a hotbed of innovation (and I would be remiss if I did not add: #LongLA), this trend is also popping up in other major metros.</p><p><a href="https://www.spacious.com/">Spacious</a> turns restaurants into coworking spaces during the day when they would otherwise be closed. Recently they have also begun expanding into vacated retail spaces. <a href="https://www.workbar.com/staples/">Staples partnered with Workbar</a> to launch coworking spaces inside Staples stores throughout Massachusetts. They have the whole nine yards — even nursing rooms.</p><p>The reality is, work is no longer confined to the office. We want to be able to work from anywhere and operators of physical spaces are capitalizing on the opportunity.</p><h3><strong>The hidden problem: access control</strong></h3><p>What very few talk about is how hard it is to innovate what you do with space, for instance, if you were to operate a coworking space. There is a lot of friction that the traditional real estate model and the services around it are not set up for.</p><p>For one, access and security seem to be diametrically opposed. Your coworking space has a rotating cast of members and their guests. While you want to make it easy for your customers to sign up for your service and come and go, you also want to make sure that only those supposed to be there are in your space without placing an undue burden on your customers or your staff.</p><p>You probably also offer a variety of packages in order to entice members with different needs: a desk or a private office, daily or only a certain number of days each month, full days or a set of hours split across different days. If your coworking space is doing well, you may want to expand locations and allow your members to float between desks and meeting spaces across locations. Then what if a member does not pay? It is not straightforward to link one’s access to their billing status. Being customer-centric unfortunately often leads to either administrative hassle or lost revenues.</p><p>Our portfolio company, <a href="https://www.nexkey.com/">Nexkey</a>, has a solution for access control for the modern workplace. Nexkey can make any lock into a smart lock that you can open with your smartphone. They retrofit any mechanical locks and fob- or card-enabled locks as well as add locks to sliding doors, which allows workplaces to manage access to all their doors digitally. They have an API where you can embed Nexkey into your apps for your members, enabling a smooth billing-and-access process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*UVINlFL3CkYcChcKjvcuLA.png" /><figcaption>Nexkey allows your employees, members, and guests to open the door with their smartphone.</figcaption></figure><p>Nexkey recently came out of stealth and is unveiling their products with deployments at various coworking spaces, innovation centers, tech startups, agencies, office spaces, and even a gym. We co-led the Seed round with K9 Ventures and I am a board member. In fact, K9 has been the incubating grounds for Nexkey’s access control solution, read about <a href="http://www.k9ventures.com/blog/2018/09/26/the-future-is-keyless-introducing-nexkey/">Manu Kumar’s take here</a>.</p><p><a href="https://onepiecework.com/">OnePieceWork</a>, one of Nexkey’s customers, only has two staff members at each location who used to spend most of their days letting people in and out. With Nexkey, access control is now integrated into their membership app and is on auto-pilot. Plus, as it is now a lot easier to enforce usage of the meeting rooms to just those who pay, they are seeing an increase in revenues.</p><p>There are security benefits as well. The operations manager can revoke access instantly, without having to wait for your former customers or visitors to return their keys. While keys or key cards can be shared, a standard Nexkey user cannot share his or her digital key. By definitively knowing who has access to their space, Nexkey gives the owners a peace of mind.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F4uoH8JAK_GY%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D4uoH8JAK_GY&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F4uoH8JAK_GY%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/44cfde333d652590cc5b87e7464e09b1/href">https://medium.com/media/44cfde333d652590cc5b87e7464e09b1/href</a></iframe><h3><strong>The future workplace</strong></h3><p>In a world where Amazon can deliver anything to you with a few taps on your phone, we tend to think that people do not go out anymore. But people do; they are just looking for different experiences. Soon, spaces that only have a single purpose will begin to seem transactional and utilitarian. We will instead talk about and spend more of our time in as well as dollars at spaces that are multipurpose and multidimensional. Spaces that embrace their clientele and build a community have more reasons to draw in customers, build emotional connections, and an enduring brand.</p><p>The world is becoming more fluid and people want to be able to seamlessly move between different spaces. As real estate innovates, we will need to rethink the operational model so that it becomes more user-friendly and intelligent.</p><p>The future workplace is poised for change and Nexkey is opening the door.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5fdb83daecc3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/working-but-not-at-work-the-future-workplace-and-nexkey-5fdb83daecc3">Working but Not at Work: The Future Workplace and Nexkey</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cameras and Cats — Five Trends That Make Computer Vision Interesting Now]]></title>
            <link>https://medium.com/upfront-insights/cameras-and-cats-five-trends-that-make-computer-vision-interesting-now-8261e60f748c?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/8261e60f748c</guid>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Tue, 25 Sep 2018 14:01:01 GMT</pubDate>
            <atom:updated>2018-09-25T15:08:50.036Z</atom:updated>
            <content:encoded><![CDATA[<h3>Cameras and Cats — Five Trends That Make Computer Vision Interesting Now</h3><h4>Why is computer vision interesting now? To answer that question, we’re going to talk about… cats.</h4><p>The internet is full of pictures of cats and cats have done a great service for computer vision. I’m recognizing their contribution.</p><p>What is computer vision and why should we care? <a href="https://insights.upfront.com/you-are-using-computer-vision-every-day-but-you-dont-know-it-yet-8898ec9b4071">Because it’s in your everyday lives and you probably don’t even know it.</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*1kE26OS9ijak8eJs2P_znw.jpeg" /></figure><p>There are five trends in this picture — five reasons that computer vision is exploding now:</p><ol><li>Cameras and sensors</li><li>Connectivity</li><li>Data</li><li>Intelligence</li><li>Computing power.</li></ol><p>I’ll tell you about them through cats. We laugh about it, but it’s incredibly hard to do and you’ll see why.</p><h3><strong>Cameras and sensors</strong></h3><p>Cameras and sensors are being deployed everywhere. Because they are better, smaller, and cheaper. There are more and more inputs of the physical world to computers. This means that there are many many cameras to take photos of cats. They are in your home outside your door, on the drones that you fly, and on the manufacturing floor to detect defects. They’re in your car on the dashboard, behind the rear-view mirror, on the sides, and at the back. In fact, your Tesla has 8 cameras, 12 ultrasonic sensors, and radar to help it see its surroundings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*_KfXbGqO0FrcLhuSqELl3Q.png" /></figure><p>We have also hugely benefitted from the rise of smartphones over the past decade. This put a camera in everyone’s pocket. Plus, the massive scale has made cameras and sensors a lot cheaper. I still remember just over 10 years ago, I had the coolest phone back then, which was a Motorola RAZR. It might have had a camera but it was so bad. Nowadays, every smartphone is able to not only take beautiful high-def photos, but also capture stunning video, measure acceleration, fix geographic location, and pretty much do anything you want. How the world has changed! While your iPhone hasn’t gotten much more expensive, the camera and the sensors that it packs have improved significantly over the generations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*vJGa86vE-bkZ7DEqXgeBJw.png" /></figure><h3><strong>Connectivity</strong></h3><p>These cameras and sensors are connected. The cameras take photos at the “edge” and then they upload to the powerful servers in the cloud to aggregate all the data and access powerful computing resources. This all helps it to constantly improve how it understands the world. Each device on the edge also shares data with each other, which allows them to work together.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*zCHfj5RRny2r7cQUkttw3A.png" /></figure><p><a href="https://www.inviarobotics.com/">inVia Robotics</a> as illustrated below makes robots for fulfillment warehouses for e-commerce. The robots pull the bins from the shelves and take them to human pickers, who then would pick the item and packs and ships. The robots navigate the warehouse autonomously by reading QR codes along the shelves and on the bins. Here, the robots function as a swarm. Together, they batch orders and optimize routes. If one human picker is slow, then more robots would line up behind the faster picker. The robots can also “defrag” the warehouse, move the bins that are more frequently accessed to be closer to the pickers and the bins that are less frequently accessed to be farther away. inVia robots coordinate together based on what they see through their cameras.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*la89l3xWvDE1zdmeT2uk_Q.png" /></figure><h3><strong>Data</strong></h3><p>There is more and more visual data being shared over the web (YouTube, Facebook, Instagram, Pinterest, Snap, Flickr). They are powered by cameras in everyone’s pocket and the many connected and distributed cameras and sensors. Over 300 hours of video are uploaded to YouTube every minute. Over 300 million photos are posted on Facebook every day. Over 50 million photos are posted on Instagram every day.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*x04-kjdOHOAiTnWmLqLfUA.png" /></figure><p>We have also benefitted from the autonomous car movement. The data collection first started with the Google Maps effort, where Google sent cars to drive around streets everywhere. Then many companies launched self-driving car efforts, including Waymo, Uber, Cruise, and Zoox. As a result, they have gathered a huge amount of data about road conditions, cars, pedestrians, and road signs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*hwobd-zTHrHFtH7HOnb6Fg.png" /></figure><h3><strong>Algorithms</strong></h3><p>Understanding what is in an image is staggeringly complex. Only recently have computers been able to do this. We have developed algorithms to simplify the computation so that it wouldn’t take days or months to tell you whether you took a photo of a cat. The most foundational algorithm for computer vision is convolution (<a href="https://insights.upfront.com/how-does-a-computer-know-that-this-is-a-cat-f5e2ede6f583">which I explain in this blog post</a>). This is why you often hear the words Convolutional Neural Networks, or CNNs, or ConvNets, when people discuss computer vision.</p><p>Convolution is just one algorithm. Researchers have developed many other algorithmic innovations to further reduce the complexity and help computers recognize what’s in an image. Most are pretty esoteric and related to the inner workings of statistics and models. I’ll give two more examples that are easier to grasp.</p><p>There is data augmentation, which is applying techniques to distort images slightly such that the images don’t change meaningfully, so that you have more data to train your model. For example, you can create mirror images, crop randomly, use rotation, shear, warp locally, shift the colors, and so on. Now you have eight pictures of cats when you started with just one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*pQ4a8HGK-Vr4cASCOZuIXg.png" /></figure><p>There is transfer learning, which is taking a model that someone else has trained based on millions of images of cats and applying it to your problem. You probably need to tweak it slightly because they were only using cat photos that are taken in professional lighting whereas you want to be able to recognize cats in photos you took on your phone. This is much easier than starting from scratch and training your own model.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*STYbgBlpGY0BPar3OVH8hQ.png" /></figure><p>As we develop better and better algorithms, we can recognize what’s shown in an image more accurately and quickly.</p><h3><strong>Computing power</strong></h3><p>The last piece I should touch on is computing power. I’ve explained to you why we need so much computing power and why we could always use more.</p><p>Traditionally, computing has relied on CPUs, or central processing units, that are great at taking a sequential list of instructions and executing them very quickly. Over time, we have progressed making CPUs that are smaller, better, and faster, in line with Moore’s Law.</p><p>Computer vision problems, on the other hand, are better suited for GPUs, or graphical processing units. Whereas CPUs have multiple cores, GPUs have thousands of cores that can operate in parallel and can perform a multitude of identical simple jobs simultaneously. GPUs were developed for video games where you need to render images on the screen efficiently. It turns out that is an analogous computation to image recognition, where when you apply algorithms like convolution, you want to apply it quickly all across the image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*xJeOalvjkEZpIYuu3MSjDQ.png" /></figure><p>This is why the GPU market has gotten a lot of attention of late. The market leader today in GPUs is nVidia, but the other big tech companies have all made serious moves to create their own GPUs and chips tailored to machine learning tasks, from Intel to Google to AWS.</p><p>Major cloud providers have been adding GPUs and machine-learning specific services to their offerings. The chart below shows CapEx spend by public cloud vendors. Although these figures don’t specifically break out how many nVidia GPUs they are buying every year, they show that it is an area of heavy investment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*W1NpIsUpwhlL6y79VTC-6w.png" /></figure><p>Why? As you can see, the public cloud market is expected to more than double from $31B in 2016 to $72B in 2019. We all need more computing power in order to recognize cats more quickly and accurately.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*-tZ_reUiP1EsMqJE0UBI1g.png" /></figure><p>There is a virtuous cycle of better sensors, increased connectivity, more data, better algorithms, and more computing power. This gives better predictions and makes computer vision better and better and more pervasive in our world.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*WDv4XShJvkEq0EeL9ST2pA.png" /></figure><h3>So what are cat photos to you?</h3><p>Computer vision is not a hardware-specific gimmick. This is breaking another wall that is allowing us to interact digitally and physically. It fundamentally changes how businesses operate in the physical world. Any business you have is either using computer vision or feeding computer vision.</p><p>What are the problems you have that computer vision can solve? I have a few ideas:</p><p><strong>Identity &amp; security </strong>— can your product become “smarter” if it can see exactly who or what it is interfacing with? This could be applied to a digital product like a phone FaceID or a physical product like a door</p><p><strong>E-commerce</strong> — why is it that I’m still flipping through online catalogs and imagining clothes on me, but when they arrive, the clothes invariably fit terribly or not what I had in mind?</p><p><strong>Change detection </strong>— so much of our jobs is to keep an eye on something.<br>When you’re driving and you want to switch lanes, you need to remember to check your blind spots. Why do we still have blind spots?</p><p>The parking police walk around the city all day to give tickets to illegally parked cars. Why not put cameras around the city and ticket cars automatically?</p><p>Or when you’re in a factory and making lots of electronic widgets, a machine could be out of alignment and start to make defective products. Why not have a camera stare at it and alert you if it sees something different?</p><p>A lot of diseases start with changes that are imperceptible to the untrained eye, like slight tremors in your fingers might indicate Parkinson’s, why not have a camera as a silent observer in your home with a direct line to your doctor?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*etoiNkKjIXpHOJEj7dK4cQ.png" /></figure><p>Why not use computer vision to solve these issues?</p><p>Computer vision is changing the way we interact with the world. We’d be hard pressed to find a business that this would not be relevant for.</p><p>This was adapted from a talk at San Diego Venture Summit on August 16, 2017. The slides I presented at the talk are here:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.slideshare.net%2Fslideshow%2Fembed_code%2Fkey%2FvUMWX6EYuyzxFB&amp;url=https%3A%2F%2Fwww.slideshare.net%2Fchangxu%2Fcameras-and-cats-why-computer-vision-is-changing-the-way-you-see-the-world&amp;image=https%3A%2F%2Fcdn.slidesharecdn.com%2Fss_thumbnails%2Fcamerasandcats-cvtalkforweb-180925033622-thumbnail-4.jpg%3Fcb%3D1537846815&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=slideshare" width="600" height="500" frameborder="0" scrolling="no"><a href="https://medium.com/media/15f81616a8c6104e5d7dae885dbc8541/href">https://medium.com/media/15f81616a8c6104e5d7dae885dbc8541/href</a></iframe><p>If you are interested in computer vision, check out my other posts:</p><ul><li><a href="https://insights.upfront.com/you-are-using-computer-vision-every-day-but-you-dont-know-it-yet-8898ec9b4071">You are using computer vision every day, but you don’t know it yet</a></li><li><a href="https://insights.upfront.com/how-does-a-computer-know-that-this-is-a-cat-f5e2ede6f583">How does a computer know that this is a cat?</a></li></ul><p>If you are building a startup using computer vision, I’d love to talk to you. Shoot me an email at <a href="/service/mailto:chang@upfront.com">chang@upfront.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8261e60f748c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/cameras-and-cats-five-trends-that-make-computer-vision-interesting-now-8261e60f748c">Cameras and Cats — Five Trends That Make Computer Vision Interesting Now</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Does a Computer Know That This is a Cat?]]></title>
            <link>https://medium.com/upfront-insights/how-does-a-computer-know-that-this-is-a-cat-f5e2ede6f583?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/f5e2ede6f583</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[computer-vision]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Thu, 20 Sep 2018 14:01:03 GMT</pubDate>
            <atom:updated>2018-09-25T15:13:05.261Z</atom:updated>
            <content:encoded><![CDATA[<h4>You look at this picture for a fraction of a second and you know it’s a cat. But how does a computer know?</h4><p>You probably are thinking: just feed a computer many many images of cats and voilà! The computer can figure out what’s a cat and what’s not. After all, that’s where the buzz phrases Machine Learning, Big Data, and Artificial Intelligence, all seem to come together. But have you wondered exactly how that works? You don’t need a Ph.D. — in fact, you don’t even need any math. Let me break it down for you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*DB091UjrYkCg-fSSzLnc6w.png" /></figure><p>How does it know this is a cat in front of a door…</p><p>…not a tiger and a sunset? The colors in the pictures and the striations on the animals are quite similar.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*8tdLmyjZoZ6Js1KIAFwkhA.png" /></figure><p>How does it know that this line delineates the chest of this cat against a door, when the image it sees is completely flat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*LN_sDhV9__ZXPWmtN9tcqA.png" /></figure><p>Plus, the computer has never seen this particular cat in this pose with this background, because you just took this photo!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*th-xJ38DL6nM1DmgSXDNxw.png" /></figure><p>And, that is, if you’re lucky enough to get the whole cat, you might just get part of a cat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*BwUjrZYCVS_t4FdGp3Ifdw.png" /></figure><h3>How does a computer know that this is a cat?</h3><p>Understanding images is incredibly hard, we take it for granted because our phone opens when it sees our face.</p><p>To a computer, an image is a collection of pixels. Each pixel is represented with three numbers from 0 to 256 that are their RGB values, or how much Red, Green, and Blue is in each pixel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*8Osr6AqB_Z7BZ9LPe3faBw.png" /></figure><p>Each photo you take on your iPhone X is 12 megapixels, which means that it has 12 million pixels.</p><p>As each pixel is represented by three numbers, each photo is represented by 36 <em>million</em> numbers. To even know it’s a cat, your computer needs to analyze 36 million numbers.</p><p>It’s actually even <em>more</em> complicated than that. Because just one pixel doesn’t tell you that it’s a cat. Groups of pixels together give you the eyes, ears, whiskers, and now you know it’s a cat. The computer needs to analyze the relationships of these 36 million numbers to each other. This means that computers need to analyze <em>trillions</em> of relationships just to look at a single image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*K21g7HUlUJFG55HE9qO25A.png" /></figure><p>So how can a computer be able to tell you if this is a cat <em>in a split second</em>, as quickly as a human can? As you can see, understanding what is in an image is staggeringly complex.</p><p>Only recently have computers been able to do this. This is where algorithms, or intelligence, come in. We have developed algorithms to simplify the computation so that it wouldn’t take days or months to tell you if you took a photo of a cat.</p><h3>Convolution</h3><p>I’m going to explain one algorithmic innovation to you that is critical to understanding images, which is convolution. It is so foundational to image recognition that you’ll be glad that you can understand it. Plus, after this bit, you can throw around the term “Convolutional Neural Networks” like it’s nobody’s business.</p><p>If you look at an image, you’ll notice two things:</p><p>One. You look at each area separately. The door ledge on the bottom right is not relevant to the fact that there are cat ears on the top.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*Vo7FSxa6RHdASOXxef6w0A.png" /></figure><p>Two. If you know what cat ears look like, then you can find all occurrences of cat ears in the entire image.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*9XdgooLBF2g2dQg7TBU-eA.png" /></figure><p>Convolution is a method that allows you to easily and quickly do those both of those things to images. Otherwise, as mentioned before, you might look at how each pixel is related to every other pixel in the entire image, or decipher each time whether you’re looking at cat ears. Using convolution saves you a lot of work.</p><p>How does it do that? Convolution takes an image, which, if you remember, is a matrix of 36 million numbers, and apply an operation to each number that takes into account of the numbers immediately surrounding it but not the numbers that are farther away.</p><p>So if we zoom in on a cat ear, I take the pixels in this small box and I <em>convolve</em> all the numbers in these pixels to come up with a new number. Now this number has the information from itself and the eight pixels surrounding it. But it doesn’t have any information from the pixels that are far away and not adjacent to it, which makes this an efficient operation. This operation is called a convolution. I <em>convolve</em> the pixels and it tells me that I’m looking at a line at a certain angle.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/320/1*FtEMM8QHwRqJQ7vqm7NyHQ.png" /></figure><p>I <em>convolve</em> over larger and larger areas and I see two lines at these angles with a furry texture in the middle. This tells me that I’m looking at the ear of a cat.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/320/1*a2eHmygwHf1fWmmJF8PgcQ.png" /></figure><p>I <em>convolve</em> all over the image and now I see that there are two cat ears, eyes, whiskers, and paws. Now I have pretty good confidence that I’m looking at a cat. This is why you often hear the words Convolutional Neural Networks, or CNNs, or ConvNets, when you hear discussions of computer vision.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dMzLVXKF4rPsRnxDcTvoPA.png" /></figure><p>There we go! Now when you see an image, you see the 1’s and 0’s (actually numbers from 0 to 256) same as what computers see. Plus, now you have a sense of the computational weightlifting they do to understand what’s in an image.</p><p>If you are interested in computer vision, check out my other posts:</p><ul><li><a href="https://medium.com/p/8898ec9b4071/edit">You are using computer vision every day, but you don’t know it yet</a></li><li><a href="https://insights.upfront.com/cameras-and-cats-five-trends-that-make-computer-vision-interesting-now-8261e60f748c">Cameras and cats — five trends that make computer vision interesting now</a></li></ul><p>If you are building a startup using computer vision, I’d love to talk to you. Shoot me an email at <a href="/service/mailto:chang@upfront.com">chang@upfront.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f5e2ede6f583" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/how-does-a-computer-know-that-this-is-a-cat-f5e2ede6f583">How Does a Computer Know That This is a Cat?</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[You Are Using Computer Vision Every Day, But You Don’t Know it Yet]]></title>
            <link>https://medium.com/upfront-insights/you-are-using-computer-vision-every-day-but-you-dont-know-it-yet-8898ec9b4071?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/8898ec9b4071</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[computer-vision]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Tue, 18 Sep 2018 14:01:01 GMT</pubDate>
            <atom:updated>2018-09-25T15:14:48.668Z</atom:updated>
            <content:encoded><![CDATA[<h4>You might have heard of the buzz phrase “computer vision” and think that it’s some futuristic thing, but it’s in your life already.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*1HkGOUwgBdMjnUk9psS2Ug.jpeg" /><figcaption>Camera Collection, by photographer Jim Golden</figcaption></figure><p>When you wake up in the morning, the very first thing you do is to authenticate into your iPhone using Face ID, which your phone can do using its front-facing camera.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*nkJkx1nDng-2Hatqx_OKwA.png" /></figure><p>While you’re still in bed, you open up your <a href="https://www.nanit.com/">Nanit</a> app, which gets its information from the camera above your baby’s crib. Nanit gives you a report on how well your baby slept last night. You see that it took Lily 12 minutes to fall asleep. She slept for 10 and half hours, and you and your husband only had to visit her twice in the middle of the night. Pretty good!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*ruN07K2CnHWNIbTZkahzGg.png" /></figure><p>As you flip through your email, you get a notification from <a href="https://waldophotos.com/">Waldo Photos</a> that they have new photos of your daughter, who’s away at camp. As teachers at camp upload the photos they take of the campers, Waldo’s image recognition works in the background to pick out just the photos including your daughter, saving you the trouble of slogging through everybody’s photos.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*FF38DZH__KYEBeTrqTyXSg.png" /></figure><p>Then as you are making breakfast, your son is playing with the iPad on the table. The iPad can see what’s on the table because it has an <a href="https://playosmo.com/en/">Osmo</a> camera on top. Your son puts together Lego-like puzzle pieces and makes the character on the screen of his iPad jump through obstacles, telling it to run right three steps, run down two steps, then jump.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*mdexWum07xJiBPh9PQaKTQ.png" /></figure><p>Your doorbell rings and you open up the <a href="https://ring.com/">Ring</a> app to see that it’s the delivery guy with a package for you. You tell him to leave the package there because your hands are full right now and you’ll pick it up later.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*6OJdzw6FcK9sZpq77GQQyA.png" /></figure><p>You finally get to work and as you go into a conference room for your meeting, the <a href="https://www.density.io/">Density</a> device installed above the door counted the number of people in the room. Density devices gather space utilization data in this room and in every room in your building. Density reports the data to your facility manager, who can then better plan the spaces in all of their buildings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*d48O9Uxo-VA08oxvy77vHA.png" /></figure><p>After work, you drive to the grocery store. You didn’t pay attention for a second and you deviated from your lane, so your car beeped at you. Because your car’s rear-view mirror has a camera on the back of it that can see the lane markings in front and can detect it when you veer away from your lane.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*2Bq2gln6K6sfVgUIh2v2aA.png" /></figure><p>You got to an <a href="https://www.density.io/">Amazon Go</a> store to pick up groceries. You simply pick up what you want from the shelves and put it in your bag. The dense mass of cameras on the ceiling and in the shelves identified you and the item you picked up. As you leave the store, you got a bill sent directly to your Amazon app.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*6dTxmhFdGUKPeZcNs9E_1Q.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*_Og-s82vXQIghparCk5FrQ.png" /></figure><p>By the way, the lettuce that you bought at the store has been grown in a greenhouse with <a href="https://lunapowered.com/">cameras</a> installed on the ceiling. These cameras monitor the crops and if it sees any discoloration on the leaves that indicated certain diseases, it would alert the farmer to treat it immediately.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*TSRCKY6iv56pwYOVXiAyGg.png" /></figure><p>While you were in the store’s parking lot, <a href="https://orbitalinsight.com/">satellites</a> orbiting the earth took a picture of the parking lot and counted the number of cars in it. They compared it with the day before, the month before and the year before. This data is then sent to a hedge fund. The hedge fund used it to make predictions on how this retailer is doing and traded based on that information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*pjbQt1JNv6e8Zl5F8U9_NA.png" /></figure><p>In the evening, you turn on ESPN and you watch the Cleveland Cavaliers play the Golden State Warriors. Little do you know, in the background, <a href="https://gumgum.com/">GumGum</a> is quietly observing the logos of sponsors on-screen, whether on the court, on the sides, or on players’ jerseys. GumGum measures the logos by their clarity, prominence, visibility, size, share of voice, and placement in order to quantify the value each sponsor received during this game. The sponsors can then use this information to calculate their ROI as well as to support their negotiations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*7_3jUwjpKUzxGFCbAvO9qw.png" /></figure><p>Everything you are doing uses computer vision. You just haven’t thought about it in any of these ways. One of the reasons that computer vision is so useful is because it is pervasive.</p><h3>What is computer vision?</h3><p>Let’s take a quick step back. You don’t need to understand the math. You just need to understand that —</p><blockquote>Computer vision is being able to interpret the physical world through cameras and sensors.</blockquote><p>In the past, we have to manually input instructions to tell machines to do one action and then another.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*RQomhBXnju9ghFWukrtQhw.png" /></figure><p>Then, computers gained the ability to infer the world through data such as Tweets, posts, and purchases. But they only understood numbers and words.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*CdHRi_Xl3rJq15w28LU2rA.png" /></figure><p>Now, computers can understand what it is <em>seeing</em> and sensing. They can take inputs that are images and videos, which allows them to <em>directly observe the physical world</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*dwwcSL-vlKPm-_GBUfG13Q.png" /></figure><p>This means that we can have many many cameras that humans don’t need to look at. Using cameras as their eyes, computers will become more knowledgeable and aware of the physical world and be able to directly interact with us.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*rd9gzWmlcNmypATA2jOeJw.png" /><figcaption>Scene from Big Hero 6. Baymax must have great computer vision hookups.</figcaption></figure><p>If you are interested in computer vision, check out my other post:</p><ul><li><a href="https://insights.upfront.com/how-does-a-computer-know-that-this-is-a-cat-f5e2ede6f583">How does a computer know that this is a cat?</a></li><li><a href="https://insights.upfront.com/cameras-and-cats-five-trends-that-make-computer-vision-interesting-now-8261e60f748c">Cameras and cats — five trends that make computer vision interesting now</a></li></ul><p>If you are building a startup using computer vision, I’d love to talk to you. Shoot me an email at <a href="/service/mailto:chang@upfront.com">chang@upfront.com</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8898ec9b4071" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/you-are-using-computer-vision-every-day-but-you-dont-know-it-yet-8898ec9b4071">You Are Using Computer Vision Every Day, But You Don’t Know it Yet</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hiring New Associates at Upfront]]></title>
            <link>https://medium.com/upfront-insights/hiring-new-associates-at-upfront-fd6e07c253f8?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/fd6e07c253f8</guid>
            <category><![CDATA[longla]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Mon, 16 Jul 2018 22:35:03 GMT</pubDate>
            <atom:updated>2018-09-17T05:31:56.550Z</atom:updated>
            <content:encoded><![CDATA[<p>We are excited to announce that we are hiring two Associates at Upfront Ventures to join our team in Santa Monica, CA. If you are interested in working closely with our partners, portfolio and the wider LA tech ecosystem, read on.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hkZMPZWAgMHJ_i8Np6dEzQ.jpeg" /><figcaption>Aerial view of our office in downtown Santa Monica, provided by our portfolio company <a href="https://www.dronebase.com/">DroneBase</a></figcaption></figure><p>This is an opportunity for you to gain invaluable experience of the venture capital industry working with a team that brings a diverse set of experiences as former entrepreneurs, operators, and now investors. You will learn about venture investing from the frontlines, help assess new investment markets and opportunities, and spend time with talented entrepreneurs as they build their businesses.</p><p>The Associate role is a two year program designed to give you unique exposure to finance, technology, and startups. You will gain relevant experience across all of these sectors, while building your network to be well-positioned for the next step in your career. Recent Upfront Associates have joined portfolio companies in management positions, launched their own startups, and continued pursuing an investing career at other funds.</p><h4>The key responsibilities for an Upfront Associate include:</h4><ul><li><strong><em>Market Research and Deal Diligence</em>:</strong> Helping identify and diligence investment opportunities, including meeting with entrepreneurs, understanding existing and new markets, analyzing business and financial models, and conducting reference checks</li><li><strong><em>Deal Execution and Analysis</em>:</strong> Executing new and follow-on financings, including formulating investment memos, structuring deals, modeling cap table scenarios, and managing legal processes through to close</li><li><strong><em>Portfolio Support</em>:</strong> Attending board meetings and working with portfolio companies on key issues, including helping them with critical strategic and financial analyses</li><li><strong><em>Community Engagement</em>:</strong> Representing Upfront at local events, bringing new entrepreneurs into the firm’s network, and generally keeping a pulse on the LA tech community</li><li><strong><em>Platform Building</em>:</strong> Supporting the wider Upfront platform and brand, including creating industry research, assisting in putting on firm events such as the Upfront Summit, and helping manage internal processes</li></ul><h4>The ideal candidate will have the following qualifications and attributes:</h4><ul><li>3–5 years of consulting / banking experience and / or operations experience in a management position at a startup</li><li>Demonstrated strong quantitative and analytical skills; can navigate financial models and large datasets to distill the key points and tell a story</li><li>Passionate about tech-enabled businesses and all things startups</li><li>Enjoys building relationships and networking</li><li>Proactive and a self-starter, high energy and a strong work ethic, have a point of view, innovative thinker and outstanding communicator</li><li>Lives in LA and has a reasonable commute to get to our office in Santa Monica</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FZtrZocDKo6yxAHYEax1pA.jpeg" /><figcaption>Our team at the 2018 Upfront Summit (missing a couple folks)</figcaption></figure><h3><strong>Interested in applying?</strong></h3><p>If so, please fill out our short application form <a href="https://goo.gl/forms/m9JbVb8iAeT9xQBC3">here</a>. We are excited to hear from you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fd6e07c253f8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/hiring-new-associates-at-upfront-fd6e07c253f8">Hiring New Associates at Upfront</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Future of Manufacturing and Hardware is Happening Today — in Shenzhen]]></title>
            <link>https://medium.com/upfront-insights/the-future-of-manufacturing-and-hardware-is-happening-today-in-shenzhen-a8f4c4d4a9f8?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/a8f4c4d4a9f8</guid>
            <category><![CDATA[supply-chain]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[manufacturing]]></category>
            <category><![CDATA[hardware]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Tue, 01 May 2018 14:01:01 GMT</pubDate>
            <atom:updated>2018-05-01T21:37:23.082Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*7UMKeA26hy2jMkLyTteb6A.jpeg" /></figure><h4>If you are in hardware, all roads lead to Shenzhen.</h4><p>In the US, we lament about how it’s hard to back hardware startups; it’s not uncommon to hear VC’s exclaim, “we don’t do hardware.” This is because an overwhelming majority of hardware startups take much longer and end up requiring much more capital to come to fruition than what entrepreneurs or investors would expect. Hardware startups are costlier on every dimension — building a prototype, setting up manufacturing, fixing quality issues, or iterating on product.</p><p>So why is Shenzhen so far ahead of the US? Shenzhen used to be an unassuming market town with a population of 30,000. In 1980, the Chinese government designated Shenzhen as the first Special Economic Zone and it has since changed rapidly. Today, the city boasts a population of 20 million, many of whom are migrant workers that power this manufacturing powerhouse. A number of major high tech companies also call Shenzhen home, including ZTE, Huawei, and DJI —they manufacture hardware, unsurprisingly.</p><p>A month ago, I decided to check out Shenzhen for myself and discovered undeniable advantages.</p><h3>HAX Accelerator and the Huaqiangbei district</h3><p>I met up with Duncan Turner, the Managing Director of a hardware accelerator based in Shenzhen called the <a href="https://hax.co/">HAX Accelerator</a>, and he kindly hosted me for the day. HAX is located in the electronics district, Huaqiangbei. I was instantly surrounded by electronics malls — tall buildings for the sole purpose of selling electronic <em>stuff</em>. There were floors dedicated to selling cables, capacitors, circuit boards, surveillance cameras, computer peripherals, drones, and a dizzying array of electronic gadgets. The components are production grade and most vendors are directly affiliated with factories in and around Shenzhen, so they mean serious business.</p><p><a href="https://shift.newco.co/what-50-buys-you-at-huaqiangbei-the-worlds-most-fascinating-electronics-market-f0384d9fca32">Reportedly</a>, you can even find unofficial iDevice repair manuals, with detailed documentation of every component of a given Apple device and how to replace and repair anything. I wonder where they got them.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*twr0ygTgPWC1KzS7UeK4jA.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*_Kjpp611koq5-4j_gpJ4zg.jpeg" /><figcaption>Electronic malls in Shenzhen</figcaption></figure><p>HAX invests in 45 hardware startups each year and supports them in every way possible: prototyping, industrial design, supply chain, Kickstarter campaigns, marketing videos, you name it. I had the pleasure to meet with a few teams and I was instantly floored — they were able to make so much progress in a short amount of time and with little financial resources. The production prototypes looked sleek and unit economics even at minimal production volume were incredible compared to the metrics I’m used to seeing in the US.</p><p>Why? Because startups could just go downstairs to the malls for most components and return to their labs to quickly build their prototypes. When they are ready for production, they can work directly with local factories and resolve issues quickly, without time zone or language barriers. Iteration cycles of months and days get reduced to hours and minutes in Shenzhen.</p><h3>From AR glasses to smart coffee machines</h3><p>I met Aaron Rowley, cofounder of <a href="https://www.enjoyvue.com/">Vue Smart Glasses</a>, and he gave me one of his AR glasses to try. Their a-ha insight: AR doesn’t have to mean vision, it could be audio, i.e., bone conduction audio built into the arms of their glasses. When Aaron’s cofounder called, he didn’t need to physically pick up his phone and could hear through his Vue glasses. He showed me that you can also tap specific patterns on the right arm of the glasses in order to hear the number of steps you’ve walked during the day. Incredible.</p><p>Another team made a coffee maker that can make coffee just the way you like it, called <a href="https://www.orendacoffee.com/">Orenda Coffee</a>. You can start with any coffee bean. Brew a cup and tell the Orenda app any changes you’d like to make: more bitter, stronger, more of a kick, and so on. The coffee maker will then adjust all the possible parameters like an expert barista — grinding method, water temperature, coffee-to-water ratio, brew time — in order to make you the perfect cup of coffee. Where have you been all my life?</p><p>“In the past two years, we’ve taken five weekends off.” One of the cofounders proudly announced. Maybe that’s part of the secret.</p><h3>*The* ecosystem for hardware and manufacturing</h3><p>In all seriousness, startups in Shenzhen are able to get so far and so quickly. It’s the value of the ecosystem. The entire package would be hard to replicate here in the US, with our differential wages, labor force, and a host of other reasons.</p><p>However, as we are building and backing companies that we hope eventually will go global, we should leverage the best and most cost-effective resources and talent available around the world.</p><p>How would it work in practice? If you are building a hardware startup, here are some ideas:</p><ol><li>If you’re just starting, spend at least a day, if not a week, a month, or a year in Shenzhen. You’ll find that you can make progress in leaps and bounds.</li><li>If you’re prototyping and iterating quickly, consider locating your engineering and R&amp;D team in Shenzhen. Ironically, some founders I met with were not sure they’ll ever hire an US-based engineering team even as they plan on moving the company to the US, due to cost.</li><li>If you’re starting or scaling manufacturing and you’re looking for a supply chain/ops manager, perhaps source talent based in Shenzhen. They can be on the ground working with local factories day-to-day vs. flying engineers back-and-forth. Plus, compensation benchmarks are lower.</li></ol><p>One thing to note if you do visit: do <em>not</em> jaywalk in Shenzhen. The surveillance cameras on the streets can recognize who you are and <a href="http://www.scmp.com/tech/china-tech/article/2138960/jaywalkers-under-surveillance-shenzhen-soon-be-punished-text">text you a ticket to your WeChat account</a> (and you can also pay your ticket in WeChat as well). What did I tell you? The whole city is on the cutting edge of hardware-software innovation.</p><p>If you want to read more about my thoughts on manufacturing and supply chain, <a href="https://insights.upfront.com/@changxu">here</a> are my other posts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a8f4c4d4a9f8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/the-future-of-manufacturing-and-hardware-is-happening-today-in-shenzhen-a8f4c4d4a9f8">The Future of Manufacturing and Hardware is Happening Today — in Shenzhen</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Startups: Don’t Just Focus on the AI]]></title>
            <link>https://medium.com/upfront-insights/ai-startups-dont-just-focus-on-the-ai-e6a294470020?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/e6a294470020</guid>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Tue, 09 Jan 2018 15:16:02 GMT</pubDate>
            <atom:updated>2018-01-09T15:16:02.742Z</atom:updated>
            <content:encoded><![CDATA[<h4>Nowadays it’s hard to come by a pitch deck without the phrases “AI” or “ML,” as if they are magic bullets. Unfortunately, the quality isn’t very high most of the time.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*1ZnxvPs09hh65bkiYzsydA.jpeg" /><figcaption>Illustration: Colin Anderson/Getty Images</figcaption></figure><p>We have several portfolio companies that use AI/ML to power their products — for example, <a href="http://qordoba.com/">Qordoba</a> uses ML to suggest translations for words, <a href="http://imbellus.com/">Imbellus</a> uses AI to generate art for their games, and <a href="https://www.gumgum.com/">GumGum</a> uses ML to identify images with specific features in order to place relevant ads — but they don’t bill themselves purely as AI companies. What’s going on here?</p><blockquote>If you are working on your pitch right now, here’s a pro tip: don’t just focus on the AI.</blockquote><blockquote><strong>Instead, focus on your business model, the opportunity you’re addressing, your go-to-market, and so on, like any other startup.</strong></blockquote><h3>Good ML, poor business case</h3><p>I’ve seen too many examples where the ML is fine and technically robust, but the business case hasn’t been thought through. For example, one startup picked where to focus based on where the data was easy to come by and more structured, without thinking about the competition (plenty) and differentiation (just ML is not enough).</p><p>Another startup applied ML in such an avant-garde fashion that only the most enlightened and forward-thinking customers would understand the product. In enterprise sales, it’s often better to cater to the common denominator than to hope that the industry will advance rapidly.</p><p>Yet another one focused their time on building a product to solve a problem, only to discover that their intended customers don’t have the ability to pay. They then realized that monetizable customers do exist but those are at least one-step removed, which complicates their go-to-market.</p><p>Finally, some startups are great extensions of university research, have great talent, and probably are on their way to making fundamental breakthroughs, but they are far from having a business yet. The VC model is not built for funding research.</p><h3>Great business case, poor ML</h3><p>Of course, companies also err the other way: great business case, poor ML. In fact, when the business case sounds too good to be true, it is often correlated with a limited underlying technology that is far from realizing the product promise. Building a regression is not ML. Setting up a rule-based Q&amp;A engine does not make a conversational agent. In today’s competitive environment for AI talent, I’m hard pressed to imagine how companies without AI DNA on the founding team can attract the necessary talent to build the products they envision.</p><p>Some of the best and incredibly powerful applications for ML are where the product evolution is so natural that it’s obvious in hindsight: <a href="https://t.co/GQbhoKP3Ui">Google Maps has steadfastly built a moat such that the product is undisputedly best-in-class</a>. Business press, to its detriment, is full of rhetoric making lofty promises about AI, but real, incredible products powered by ML are built from adding up incremental, everyday investments.</p><h3>Solid business case, great ML</h3><p>The magical combination is a solid business case supported by an astute application of AI. AI is an enabling technology. Here are some questions I ask when I evaluate AI startups:</p><ul><li>Does it leverage a large dataset? Ideally, the said dataset is not in places where Google, Amazon, or Facebook would have even larger datasets.</li><li>Has the AI technology just reached or is approaching the tipping point between research and becoming useful in real-life applications? For example, image recognition and autonomous driving are relatively mature, while there has been rapid recent advances in NLP.</li><li>Does it solve an immediate need? Is it a killer app? Can it solve the cold start problem so that there is a customer at the ready and they are willing to pay?</li></ul><p><em>✌ I am an early stage investor at Upfront Ventures. If you’re working on AI/ML, drop me a line at </em><a href="/service/mailto:chang@upfront.com"><em>chang@upfront.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e6a294470020" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/ai-startups-dont-just-focus-on-the-ai-e6a294470020">AI Startups: Don’t Just Focus on the AI</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Future of Work? Rebranding EdTech]]></title>
            <link>https://medium.com/upfront-insights/future-of-work-rebranding-edtech-7c1423680fc2?source=rss-84ada277731c------2</link>
            <guid isPermaLink="false">https://medium.com/p/7c1423680fc2</guid>
            <category><![CDATA[education]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[edtech]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Chang Xu]]></dc:creator>
            <pubDate>Tue, 31 Oct 2017 14:01:02 GMT</pubDate>
            <atom:updated>2017-10-31T14:01:02.162Z</atom:updated>
            <content:encoded><![CDATA[<p>“What is education anyway but training for work and life?” my partner <a href="https://medium.com/@karanortman">Kara Nortman</a> mused. Nevertheless, EdTech conjures a totally different brand image (old, low multiples, slow moving, small exits) than the new buzzword, “Future of Work” (big, top of mind given the election, fits macro shifts around AI and robots). They are more similar than not.</p><p>Ever since I co-founded an EdTech startup five years ago, I’ve fallen in love with this space. I saw that a good education has outsized impact compared to other interventions at uplifting individuals and transforming societies. What’s more, I believe that this is the moment in time where technology can catalyze this impact, regardless of what term is in vogue.</p><h3>EdTech has a branding problem</h3><p>People define education too narrowly.</p><p>When I talk to most people about EdTech, they think about software that sells to K-12 or higher ed institutions. They see a decentralized system of schools that all purchase independently, lacking a standardized curriculum, catering to strong teacher unions, and being muddled in public policy debates. A headache, indeed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*OadgkutUHPutzJY5WEmIFg.jpeg" /><figcaption>Diamondale Elementary School, a typical K-12 setting</figcaption></figure><p>As a result, many VCs state flatly, “I don’t do EdTech” — though they give philanthropically to education nonprofits.</p><p>Many entrepreneurs also shy away from EdTech — though they tell me that after they’ve made money, they would love to make an impact in education.</p><p><strong>I believe there are a number of venture-scale businesses to be built in EdTech that are transformative in both their impact and financial outcomes.</strong> And, now is the time to build them. Let me tell you why.</p><h3>What most investors don’t know: EdTech is exciting!</h3><p>We think of education as confined by four walls and something that one does between the ages of 6 and 18 (or maybe 22).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*Ute7mcbOXq8-GRviQ9K6bw.png" /></figure><p>But for children and youths, there are many more opportunities outside of the classroom where learning happens. Look at the myriad of toys, books, activities, or anything in the world. If schools facilitate formal learning, then these are conduits for informal learning.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*rMs5LFX0SXMoGCF0UFP7AA.png" /></figure><p>Similarly, for adults, there are many opportunities to continue learning and adapting for career advancement and for leisure. They are corporate training, coding bootcamps, MOOCs, podcasts, conferences, and fun classes such as cooking and fitness. They surround our work and life.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*aZ0WLHIfWOxwJ-bp-y3Y9Q.png" /></figure><p>We should define education more broadly to encompass any opportunity where learning could happen and where people could do something now that they couldn’t do before.</p><p><strong>Education, broadly defined = Learning + Talent</strong></p><h3>Change will start at the periphery</h3><p>It’s true: innovations targeting formal learning struggle with distribution and, therefore, viability. Although the overall spend on education is huge, in the US, it’s difficult to access the vast majority of the funds for structural reasons, leading to sub-optimal addressable markets.</p><p>But there is certainly no shortage of problems (and opportunity): <a href="http://www.washingtonexaminer.com/education-spending-skyrockets-year-after-year-but-student-achievement-stays-stagnant/article/2547806">in K-12, education spending per student has more than doubled over the past three decades, but outcomes haven’t budged</a>. In higher ed, rapidly rising tuition costs have led <a href="https://studentloans.net/student-loan-debt-statistics/">student debt burden to balloon to $1.4T nationally</a>, with <a href="http://time.com/money/4701506/student-loan-defaults-record-2016/">one in six in default</a>. <a href="https://www.theatlantic.com/education/archive/2016/04/the-slow-growth-of-college-graduation-rates/477798/">College completion rate is at less than 50% </a>— and most Americans don’t even go in the first place.</p><p>My bet is that <strong>disruptive innovation and transformative businesses in education will start on the periphery, tackling informal learning</strong>. Eventually they will influence the core of formal learning. Here are some examples of what this looks like.</p><h3>Maker toys are hiding in plain sight</h3><p>For example,<strong> </strong>I love <strong>children’s toys that look like candy to kids and vegetables to parents.</strong></p><p>Traditional toys are built by toymakers who optimize for fun, while educational products are built by education folks who care mostly about teaching arithmetic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*vH-z3Q3YndPEiomgYce4Ug.png" /><figcaption>Left: Mr. Potato Head is a classic toy. Right: Math problems thinly disguised by a game</figcaption></figure><p>We are finally seeing a convergence of the two worlds. “Maker” toys such as <a href="https://www.youtube.com/watch?v=rfQqh7iCcOU">Makey Makey</a>, <a href="https://medium.com/@littleBits">LittleBits</a>, and <a href="https://medium.com/sam-labs">SAM Labs</a> are not only addictive experiences, but also are the best conduits for learning STEM concepts around electricity. <strong>Learning STEM through playing with these toys is way better than using textbooks</strong>, because to create is to demonstrate the highest level of mastery.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*QstIM-5NBRxYSVkORtSxkA.jpeg" /><figcaption>“Banana piano”: using bananas to conduct electricity with Makey Makey</figcaption></figure><p>Furthermore, there is a real market for selling toys directly to consumers. The toys market is <a href="http://www.toyassociation.org/TIA/Industry_Facts/salesdata/IndustryFacts/Sales_Data/Sales_Data.aspx?hkey=6381a73a-ce46-4caf-8bc1-72b99567df1e">$20B in the US</a> and <a href="http://www.toyassociation.org/PressRoom2/News/2016_News/Global_Toy_Market_to_Surpass_90_Billion_in_2016.aspx">$90B globally</a>. If you can achieve a mass price point, then they are also products that everyone loves to gift, further boosting the commercial viability.</p><h3>Alternative models of higher ed are buzzing with activity</h3><p>On the other end of the spectrum, I also love efforts to democratize higher education. Move away from the monolithic four-year, full time, high cost models to <strong>learn-as-you-go, learn anywhere, affordable, scalable, alternative models</strong>.</p><p>One set of alternative models is Massively Online Open Courses, or MOOCs. Though they have had their <a href="https://www.edsurge.com/news/2017-02-22-the-flip-side-of-abysmal-mooc-completion-rates-discovering-the-most-tenacious-learners">struggles</a> <a href="http://www.slate.com/blogs/future_tense/2013/07/19/san_jose_state_suspends_udacity_online_classes_after_students_fail_final.html">initially</a>, recent efforts at iterating on the product and at reinvention are more successful. <a href="https://medium.com/ycombinator">YCombinator</a> was the latest to launch a MOOC, Startup School, that <a href="https://medium.com/the-oyster/why-y-combinator-should-be-an-inspiration-to-edtech-startups-a9af7a03bd04">helps them with lead gen</a>. <a href="https://medium.com/udacity">Udacity</a> launched their <a href="https://blitz.com/">Blitz</a> program to further help their nanodegree graduates get jobs, to a lot of demand from both graduates and employers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*UI39s1zbyYLzoGjKoDzGJQ.jpeg" /><figcaption>Student learning physics online through Udacity (credit: NPR)</figcaption></figure><p>Another set are coding bootcamps, which are typically in-person, three-month crash courses to train junior developers. Today, 18,000 students enroll per year paying ~$11,000 tuition per head. This <a href="https://www.coursereport.com/reports/2016-coding-bootcamp-market-size-research">$200M market</a> virtually didn’t exist five years ago.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*S0iw-UT_JTUIedeumytzJw.jpeg" /><figcaption>Students studying to become junior developers in a coding bootcamp</figcaption></figure><p>These examples are just scratching the surface.</p><p>Our world is changing rapidly and, with it, the skills that are relevant are ever-evolving. It is no longer possible to expect to have the same job for forty years.</p><p>Developing alternative models for lifelong learning and enabling employment will become ever more critical. They will be some of the most impactful and biggest businesses in the decades to come. This is why I am as jazzed as ever about EdTech.</p><p><em>✌ I am an early stage investor at </em><a href="https://upfront.com/"><em>Upfront Ventures</em></a><em>. I am passionate about the intersection of improving human potential and big, transformative businesses. If you care about the same things, drop me a line at chang@upfront.com.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7c1423680fc2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/upfront-insights/future-of-work-rebranding-edtech-7c1423680fc2">Future of Work? Rebranding EdTech</a> was originally published in <a href="https://medium.com/upfront-insights">Upfront Insights</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>