<?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 Josef Feldman on Medium]]></title>
        <description><![CDATA[Stories by Josef Feldman on Medium]]></description>
        <link>https://medium.com/@josef_feldman?source=rss-efa742544b48------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*Q7c23pg3-3dZaZuV.png</url>
            <title>Stories by Josef Feldman on Medium</title>
            <link>https://medium.com/@josef_feldman?source=rss-efa742544b48------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 12:24:31 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@josef_feldman/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[Autonomous Vehicles Panel Summary]]></title>
            <link>https://medium.com/venture-capital-think-tank/in-recent-years-autonomous-vehicles-have-become-a-hot-topic-for-investors-worldwide-d5dcd2045a76?source=rss-efa742544b48------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5dcd2045a76</guid>
            <category><![CDATA[autonomous-cars]]></category>
            <category><![CDATA[self-driving-cars]]></category>
            <dc:creator><![CDATA[Josef Feldman]]></dc:creator>
            <pubDate>Thu, 13 Apr 2017 22:35:16 GMT</pubDate>
            <atom:updated>2017-04-13T22:37:18.456Z</atom:updated>
            <content:encoded><![CDATA[<p>In recent years, autonomous vehicles have become a hot topic for investors worldwide. ThinkTank.vc recently hosted a panel of investors and founders to discuss the state of the market, technology trends, and challenges to watch out for. Companies represented included Toyota Research Institute, Grishin Robotics, Lux Capital, Nuro, Innoviz, Auro Robotics, Founder’s Guild, Pearl Automation and others.</p><p>To level-set, it’s worth defining the various levels of autonomy (this <a href="http://www.techrepublic.com/article/autonomous-driving-levels-0-to-5-understanding-the-differences/">article</a> does a good job).</p><ul><li><strong>Level 0</strong>: this is driving as most people know it. The human controls everything.</li><li><strong>Level 1</strong>: the driver-assistance level, where certain functions like steering or accelerating can be done by the vehicle automatically</li><li><strong>Level 2</strong>: Driver doesn’t have to have hands on steering wheel or pedals, but must be ready to take over</li><li><strong>Level 3</strong>: Drivers still necessary but most “safety-critical” functions are shifted to the vehicle</li><li><strong>Level 4</strong>: Fully-autonomous, car can drive itself</li><li><strong>Level 5</strong>: no steering wheel in the car</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/957/0*SjmzyGDN8WSg8A5t." /></figure><p><strong>Aftermarket/retrofit (incremental innovation) vs Designing for Level 5</strong></p><p>The panelists seemed to agree that increasing capabilities incrementally is the way to go as it provides a foundation for further development. It makes sense to gain experience with a lower level of autonomy and use lessons learned to grow from there. New sensors, algorithms, and systems will eventually make their way into OEMs, similar to how safety, emissions, and in-car entertainment slowly trickled their way into the mainstream.</p><p>The question that comes up relates to Hardware — there is consensus that machines are trained on sensors, but which sensor is best to use to train the algorithm? Is it better to go for expensive sensors that allow a system to grow into them, or go for cheaper ones that enable you to get to market sooner? The latter option may seem appealing but will necessitate rewriting all prior code when the “sensor suite” becomes commoditized and drops in price.</p><p>While Tesla started with Cameras and was able to achieve Level 2/3, their hardware is now limiting them from growing further and they may face a system re-design using LIDAR. In parallel, Google made a bet that LIDAR prices will drop by the time the market is ready for autonomous vehicles, and has designed for a LIDAR system from the beginning. Participants generally agreed that Google is in a better position in this race. An article featuring a more in-depth discussion of Tesla vs Google autonomous vehicles can be found <a href="https://www.allaboutcircuits.com/news/tesla-vs-google-do-lidar-sensors-belong-in-autonomous-vehicles/">here</a>.</p><p>Investors pointed out that in addition to autonomous vehicles, there will be many businesses that spring up around the autonomous economy: e.g., fleet operators, insurers, baggage handlers</p><p><strong>The Autonomous Ecosystem</strong></p><p>The autonomous market is still early and many startups are trying to position themselves for a future that should be “really big” and are leveraging VC funding until they get there. Some, like Cruise and Mobileye have already realized multi-billion-dollar outcomes, but those may be the exceptions to the rule. It’s worth noting that Mobileye took more than 15 years to build its position as a supplier to OEMs, which may be longer than some investors are willing to wait.</p><p>Startups that have a tangible business model and focus on generating revenue (e.g., Auro, Starsky, Drive.ai) will be in a good position to capitalize on the burgeoning autonomous economy. Participants gravitated towards the idea that shared mobility companies, which will operate fleets of self-driving cars, will be few and far in between: namely 2–3 (or fewer) of them would own the market. It’s also interesting to note that more niche markets outside of passenger cars may present unique investment opportunities: e.g., food delivery, materials for contractors, construction supplies.</p><p>Cities like New York, Chicago, San Francisco, and Boston are familiar with ride-sharing and are likely to see the direct impact of autonomous vehicles sooner than their rural counterparts. Urbanization will impact the transition to autonomous vehicles since many suburbs and rural areas will likely prefer to still own their cars due to convenience and cost. Furthermore, government regulation will play a big role; as regulators realize that the technology underlying autonomy has sufficiently matured and safety is guaranteed, they will be increasingly open to minimizing regulatory roadblocks and enabling mass adoption.</p><p>Investors and founders agreed that while Level 4 is feasible in the next 5 years, Level 5 will be very difficult given consumers’ low tolerance for machine-made mistakes. Industrial applications that don’t involve passengers are likely to see full autonomy quicker.</p><p>There was general optimism across participants regarding the future of autonomous vehicles, with most opportunities splitting into two buckets: enabling companies that provide systems/components (e.g., LiDAR, Radar, Computer Vision, etc.), and the service operators (i.e. the ones who actually move people, goods, and products). The major outstanding concerns relating to progress remain public sentiment (what if one company rushes to market and instills fear in the average consumer?), too much focus on improving technology without keeping in mind the broader transportation/accessibility problems, and regulations.</p><p>Join the ThinkTank.VC community to participate in these and other panels on Slack in the future, and get opportunities to interact with thought leaders in prominent technology areas.</p><p>Authored by <a href="https://www.linkedin.com/in/amalorodov/">Alex Malorodov</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5dcd2045a76" width="1" height="1" alt=""><hr><p><a href="https://medium.com/venture-capital-think-tank/in-recent-years-autonomous-vehicles-have-become-a-hot-topic-for-investors-worldwide-d5dcd2045a76">Autonomous Vehicles Panel Summary</a> was originally published in <a href="https://medium.com/venture-capital-think-tank">ThinkTank.vc</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Is AI Practical for commercial Applications?]]></title>
            <link>https://medium.com/venture-capital-think-tank/is-ai-practical-for-commercial-applications-3e0b3f62b352?source=rss-efa742544b48------2</link>
            <guid isPermaLink="false">https://medium.com/p/3e0b3f62b352</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Josef Feldman]]></dc:creator>
            <pubDate>Wed, 18 Jan 2017 02:19:32 GMT</pubDate>
            <atom:updated>2017-01-18T02:19:32.341Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Phil Boyer [1:01 PM] </em><br>Howdy everyone! I’ll be kicking off the discussion here.</p><p>Josef Feldman [1:01 PM] <br><a href="http://twitter.com/jason">@jason</a> <a href="http://twitter.com/philboyer">@philboyer</a> will be leading a Data Chat shortly.</p><p>The topic is — is AI really practical for commercial applications?</p><p>Please read: <a href="https://www.technologyreview.com/s/601139/how-google-plans-to-solve-artificial-intelligence/">https://www.technologyreview.com/s/601139/how-google-plans-to-solve-artificial-intelligence/</a></p><p>Participants include:</p><p>Dennis Mortensen- CEO and Founder, X.ai</p><p>Sandy Steier- CEO and Co-founder, 1010 Data</p><p>Naveen Selvadurai — CoFounder, FourSquare</p><p>Ann Miura-Ko — Founding Partner, Floodgate</p><p>Sheila Gulati- Managing Director, Tola Capital</p><p>Shivon Zilis- Founding Member, Bloomberg Beta</p><p>Alex White, Co-founder, Next Big Sound</p><p>Amit Karp — Vice President, Bessemer Venture Partners</p><p>Jesse Beyroutey- Partner, IA Ventures</p><p>Ross Fubini- Partner, Canaan Partners</p><p>Mark Johnson- Co-Founder and CEO, Descartes Labs</p><p>Jason Black- Analyst, RRE</p><p>Phil Boyer- Associate, Crosslink Capital</p><p>Sven Kreiss — Lead Data Scientist, Wildcard</p><p>Jim Hao- Associate, Firstmark</p><p>Morgan Polotan- Bloomberg Beta</p><p>Scott Clark- Founder, SigOpt</p><p>Sage Wohns- Founder, Agolo</p><p>Nikhil Srinivasan- Founder, Distributed Systems</p><p>Peter Brodsky- Founder, Hyper Science</p><p>Tim Porter- CEO and Founder, Gluru</p><p>Phil Boyer [1:03 PM] <br>Topic: Is AI really practical for commercial applications?</p><p>While this is the official topic, I think a good place to start is to ask a couple more specific questions: When is AI practical vs. overkill? Is it practical for all enterprises? What are some interesting areas where AI is already making a lasting impact? When is AI NOT worth the investment?</p><p>Jason Black [1:04 PM] <br><a href="http://twitter.com/philboyer">@philboyer</a>: great clarification</p><p>[1:05] <br>I think the upfront investment in AI for most enterprises doesn’t make sense at this point as AI startups fill the gaps</p><p>[1:06] <br>Especially if the enterprise isn’t already heavily leveraging data science and machine learning</p><p>[1:07] <br>And as increasingly larger companies get more comfortable with smaller and smaller software providers, it creates a huge opportunity for startups to sell into those accounts</p><p>[1:08] <br>that being said, it depends on how closely the 3rd party solutions fall to each enterprises’ core competency</p><p>Phil Boyer [1:08 PM] <br>However, Salesforce recently announced 3 acquisitions of AI companies (seemingly acquihires) in order to help develop applications. Do we think this will be a successful strategy for them? Or will the startups win?</p><p>Jason Black [1:11 PM] <br>Historically Salesforce has used acquisitions to build out their product offerings. Heroku, Desk.com, RelateIQ, Exact Target, etc. are great examples there. They haven’t done a ton to use that new talent to augment their existing products to date — though that can always change</p><p>[1:13] <br>Continuing on the earlier discussion, I feel like there are also going to be tools out there that just make it easier to build in house as well, thereby lowering the upfront investment</p><p>[1:13] <br>Tools like prediction.io, h2o.ai, or platform providers like databricks</p><p>Phil Boyer [1:14 PM] <br>Fair. I think the technologies (open source, commercial dev tools) will lower upfront investment. But do we have examples of companies or industries where the investment is showing positive ROI yet? Still too early to tell/</p><p>[1:14] <br>?</p><p>Jason Black [1:15 PM] <br>Would love to hear from <a href="http://twitter.com/peter">@peter</a> <a href="http://twitter.com/prashant">@prashant</a> <a href="http://twitter.com/sagewohns">@sagewohns</a> <a href="http://twitter.com/sheila">@sheila</a> or <a href="http://twitter.com/markjohnson">@markjohnson</a> who I see are in the caht here…</p><p>Phil Boyer [1:16 PM] <br><a href="http://twitter.com/markjohnson">@markjohnson</a> started a company on the premise that deep learning is a worthwhile investment. Would suspect he has an opinion here…</p><p>[1:17] <br>#callout</p><p>Jason Black [1:19 PM] <br>Well, I think the continued growth of data teams within enterprises is a proxy for an ROI</p><p>Josef Feldman [1:19 PM] <br>I think breaking this into target markets is helpful — enterprise &amp; Dev tools are significantly different from consumer applications.</p><p>Phil Boyer [1:20 PM] <br><a href="http://twitter.com/jason">@jason</a>: Maybe. It is definitely a proxy for “I” — not necessarily “R”!</p><p>Jason Black [1:20 PM] <br>I just spoke to an data lead at AT&amp;T who said they’re leveraging data down to the individual truck and have determined that, due to the increase in traffic accident rates and decrease in speed, AT&amp;T service trucks shouldn’t make left turns at intersections</p><p>[1:20] <br>which is pretty intense</p><p>[1:21] <br><a href="http://twitter.com/philboyer">@philboyer</a>: that’s a fair characterization</p><p>[1:21] <br>but you’d expect data teams to stagnate or shrink if there wasn’t a justifiable “R”</p><p>[1:22] <br>though it’s an interesting observation that maybe the “R” is defined not by $ but maybe by some other metric</p><p>Phil Boyer [1:22 PM] <br>Yes — and it depends on what enterprises feel is an appropriate timeline to seeing the returns on investment</p><p>Jason Black [1:22 PM] <br>just as much Android adoption doesn’t directly affect Google’s bottom line, but it does increase their likelihood that a particular user uses Chrome to search the web (and spend more time on the web in general)</p><p>[1:22] <br>also very true</p><p>[1:23] <br>I think at a certain point, though, it will be too costly *not* to leverage data and ML</p><p>[1:23] <br>and maybe the investment today is driven by that expectation</p><p>Phil Boyer [1:24 PM] <br>True. I think there is starting to be a “keeping up with the Joneses” attitude in enterprise adoption/hiring in data science / ML</p><p>Jason Black [1:24 PM] <br>exactly</p><p>Phil Boyer [1:25 PM] <br>They prob don’t know what the F they are doing though</p><p>Jason Black [1:25 PM] <br>Gartner actually recently put out a report on Hadoop that highlighted a lot of the challenges</p><p>Morgan [1:25 PM] <br>no different than adoption of software and the web in the 90’s and 00’s was “keeping up with the joneses”?</p><p>[1:25] <br>i see ML as eating software, which is eating the world</p><p>Jason Black [1:25 PM] <br>uploaded and commented on this image: gartner_hadoop-100585449-large.idge.png<br>1 Comment<br>This doesn’t bode well for Hadoop’s biggest vendors. After all, as Gartner analyst Nick Huedecker posits, “Hadoop [is] overkill for the problems the business[es surveyed] face, implying the opportunity costs of implementing Hadoop [are] too high relative to the expected benefit.”</p><p>Jason Black [1:25 PM] <br>^ source <a href="http://www.infoworld.com/article/2922720/big-data/hadoop-demand-falls-as-other-big-data-tech-rises.html">http://www.infoworld.com/article/2922720/big-data/hadoop-demand-falls-as-other-big-data-tech-rises.html</a> (edited)</p><p>[1:25] <br>ML = software</p><p>[1:26] <br>But I get what you’re saying</p><p>[1:26] <br>there are a lot of deterministic, rules based programs out ther</p><p>Phil Boyer [1:26 PM] <br><a href="http://twitter.com/morganpolotan">@morganpolotan</a>: I agree — I think that ML is an extension of where software is today and AI the next extension.</p><p>Morgan [1:26 PM] <br><a href="http://twitter.com/philboyer">@philboyer</a>: how do you differentiate ML and AI?</p><p>[1:26] <br>line always seemed blurry to me</p><p>Jason Black [1:26 PM] <br>those will likely start to leverage more subtle decision making engines enabled by AI</p><p>Phil Boyer [1:26 PM] <br>the question is really is investment into AI seeing returns yet?</p><p>[1:28] <br>The way I think of it, ML enables a program to select the appropriate algorithm to help arrive at a desired outcome. AI is essentially modeled after the brain to determine what the desired outcome is, which can constantly change based on a multitude of factors.</p><p>Jason Black [1:28 PM] <br>added and commented on this Plain Text snippet: ML v AI<br>“Machine learning deals with designing and developing algorithms to evolve behaviors based on empirical data. One key goal of machine learning is to be able to generalize from limited sets of data (paraphrased from [1]). Russell and Norvig [2] lists machine learning as a specific capability, namely the ability to “adapt to new circumstances and to detect and extrapolate patterns”.<br>Artificial intelligence encompasses other areas apart from machine learning, including knowledge representation, natural language processing/understanding, planning, robotics etc.”<br>1 Comment<br>Just to answer your question, <a href="http://twitter.com/philboyer">@philboyer</a></p><p>Morgan [1:30 PM] <br>love that definition</p><p>Sage Wohns [1:30 PM] <br>that’s a great definition <a href="http://twitter.com/jason">@jason</a>, I see AI as a ML stack that able to emulate previously human-exclusive properties</p><p>Morgan [1:30 PM] <br>people use both terms interchangeably</p><p>Sheila [1:31 PM] <br>I’ve seen a lot of companies that provide data-centric outcomes (using ML or AI) for enterprises (and govt) and the results can be extraordinary. Often better than the results we would see from an internal data science team initially. I’m very positive on ML and AI companies driving real business outcomes (and then becoming a feature of every software company ever time). I do believe we will see greater productivity of internal data science as well especially as the third party platform tech and tooling gets better. The opportunity is so vast that there is clearly room for all (bespoke solutions, horizontal and vertical AI and ML, internal data science enablement and teams)</p><p>Phil Boyer [1:36 PM] <br>Agreed <a href="http://twitter.com/sheila">@sheila</a>. However when I see a company in the space, I want them to focus on the best solution for the problem they are solving, not necessarily the technology at the forefront. Using ML or AI as your software approach seems to make the most sense when you have a massive, high quality dataset that continuously improves your application. in other cases, it may not be worth the investment in infrastructure, talent, etc.</p><p>Jason Black [1:39 PM] <br>Returning to the initial question around when does it make sense to build/leverage ML within an enterprise, there’s probably some way to categorize the stages companies take on the path to data/ML enlightenment. Maybe something along the lines of: identifying data sources, data exploration and transformation, data insights, data/ML driven decisions, formalized data/ML analysis pipeline/mechanisms, data integrated into product/core tooling, data enlightenment</p><p>[1:41] <br>Any thoughts on that path?</p><p>[1:42] <br>Feel like most non-tech enterprises are probably hovering around “data exploration and transformation”</p><p>[1:43] <br>Hence the large opportunity to step in and provide tools/products/services for things much further down the chain</p><p>Phil Boyer [1:44 PM] <br>ETL and data cleansing seems to be the area where most enterprises still get hung up. Hard to scale this appropriately when you have a lot of disparate data sources and formats</p><p>Jason Black [1:46 PM] <br>yeah, maybe it makes sense to break that out into 2 steps then: first “data transformation/cleansing” and then “data exploration”</p><p>Phil Boyer [1:47 PM] <br>Alright folks — have to sign off. Some great stuff in here! thanks for joining us</p><p>Morgan [1:47 PM] <br>great talk!</p><p>Jason Black [1:48 PM] <br>Same. Thanks, everyone!</p><p>[1:48] <br><a href="http://twitter.com/josef_feldman">@josef_feldman</a> &amp; <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a> will be in touch about the next round</p><p>Josef Feldman [1:49 PM] <br>Thanks guys! <a href="http://twitter.com/philboyer">@philboyer</a> <a href="http://twitter.com/jason">@jason</a> <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a> <a href="http://twitter.com/channel">@channel</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3e0b3f62b352" width="1" height="1" alt=""><hr><p><a href="https://medium.com/venture-capital-think-tank/is-ai-practical-for-commercial-applications-3e0b3f62b352">Is AI Practical for commercial Applications?</a> was originally published in <a href="https://medium.com/venture-capital-think-tank">ThinkTank.vc</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The role of open source Machine Learning platforms in the startup ecosystem— Conversation with…]]></title>
            <link>https://medium.com/venture-capital-think-tank/the-role-of-open-source-ml-platforms-in-the-startup-ecosystem-conversation-with-sandy-steier-911281facf36?source=rss-efa742544b48------2</link>
            <guid isPermaLink="false">https://medium.com/p/911281facf36</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[mls]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[tensorflow]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[Josef Feldman]]></dc:creator>
            <pubDate>Wed, 18 Jan 2017 02:15:27 GMT</pubDate>
            <atom:updated>2017-01-18T02:16:34.769Z</atom:updated>
            <content:encoded><![CDATA[<h3>The role of open source Machine Learning platforms in the startup ecosystem— Conversation with Sandy Steier (1010Data), Mark Johnson (Descarte Labs), and Jason Black (RRE)</h3><p><em>Jason Black [1:49 PM] </em><br>Hi everyone, you should have all been shared on the prompt and the article via email. If not, today we’ll be discussing the role of open source ML platforms on the startup ecosystem. Please read Ben Thompson’s take on TensorFlow to get the discussion rolling (<a href="https://stratechery.com/2015/tensorflow-and-monetizing-intellectual-property/">https://stratechery.com/2015/tensorflow-and-monetizing-intellectual-property/</a>) and feel free to bring your own articles for the discussion. Talk soon!</p><p>Josef Feldman [3:01 PM] <br><a href="http://twitter.com/jason">@jason</a>: will be kicking things off momentarily. Here’s a list of everyone in this group, feel free to connect directly!</p><p>Dennis Mortensen- CEO and Founder, X.ai<br>Sandy Steier- CEO and Co-founder, 1010 Data<br>Naveen Selvadurai — CoFounder, FourSquare<br>Ann Miura-Ko — Founding Partner, Floodgate<br>Sheila Gulati- Managing Director, Tola Capital<br>Shivon Zilis- Founding Member, Bloomberg Beta<br>Alex White, Co-founder, Next Big Sound<br>Amit Karp — Vice President, Bessemer Venture Partners<br>Jesse Beyroutey- Partner, IA Ventures<br>Ross Fubini- Partner, Canaan Partners<br>Mark Johnson- Co-Founder and CEO, Descartes Labs<br>Jason Black- Analyst, RRE<br>Phil Boyer- Associate, Crosslink Capital<br>Sven Kreiss — Lead Data Scientist, Wildcard<br>Jim Hao, Associate, Firstmark<br>Morgan Polotan, Bloomberg Beta</p><p>[3:01] <br><a href="http://twitter.com/morganpolotan">@morganpolotan</a>: welcome!</p><p>Morgan [3:02 PM] <br>thanks <a href="http://twitter.com/josef_feldman">@josef_feldman</a>!</p><p>Jason Black [3:03 PM] <br>Would love to kick things off with some initial comments on the article. I think the most important piece in there is Ben’s mention of the role of the “unreasonable effectiveness of data”</p><p>[3:04] <br>Very few ML companies these days bank on the fact that their algorithms are better than other companies (exceptions being DeepMind), but rather on amassing a unique data set that can be applied to a novel problem domain.</p><p>[3:04] <br>I know at Dextro — a portfolio company going after video analysis — they emphasize the role of their training data is the key value driver</p><p>Mark Johnson [3:05 PM] <br>When VCs ask me what our secret sauce is, I tell them “our dataset and platform.” Though I’d add to that list: measurement and validation (i.e., do you know that your algorithms are performing/improving/working and how can you tell?)</p><p>Jason Black [3:05 PM] <br>the open sourcing of the underlying platforms only helps them (Dextro) drive more data to their platform because they’re able to focus on business goals (edited)</p><p>Morgan [3:06 PM] <br>my favorite article on this topic is Matt Turck’s “Data Network Effects” <a href="http://mattturck.com/2016/01/04/the-power-of-data-network-effects/">http://mattturck.com/2016/01/04/the-power-of-data-network-effects/</a><br>Matt Turck<br>The Power of Data Network Effects<br>In the furiously competitive world of tech startups, where good entrepreneurs tend to think of comparable ideas around the same time and “hot spaces” get crowded quickly with well-funde…<br>Jan 4th, 2016 at 2:00 AM</p><p>Jason Black [3:06 PM] <br><a href="http://twitter.com/morganpolotan">@morganpolotan</a>: yeah, that’s a great one</p><p>[3:07] <br>And building on that: <a href="http://versionone.vc/data-not-algorithms-is-key-to-machine-learning-success/">http://versionone.vc/data-not-algorithms-is-key-to-machine-learning-success/</a><br>Version One<br>Data, not algorithms, is key to machine learning success<br>There has been an explosion in machine learning activity, and Shivon Zilis recently mapped out the current machine intelligence ecosystem as we enter 2016. This is one of the key areas that we’ll b…<br>Jan 6th, 2016 at 10:56 AM</p><p>Phil Boyer [3:08 PM] <br>Agree with Ben that open sourcing TensorFlow makes complete sense for Google, as Google’s value is in the size/quality of its data asset and the quality of its data infrastructure. Letting the crowd use and improve the quality of the ML algorithm only helps Google.</p><p>Morgan [3:08 PM] <br>since we all seem to be in agreement, is there anyone that thinks this is a bad move?</p><p>Jason Black [3:08 PM] <br>Well it helps Google, but it also happens to decrease the startup cost of building out that infrastructure if you are starting a new business</p><p>Mark Johnson [3:09 PM] <br><a href="http://twitter.com/phil">@phil</a> I argue that Google’s great asset is not just their data infrastructure, but their ability to quickly run algorithms through a testing pipeline to see what is effective — that forces a ton of work on the measurement side.</p><p>[3:09] <br><a href="http://twitter.com/morgan">@morgan</a> What does “this” refer to?</p><p>Phil Boyer [3:09 PM] <br>There are bigger hurdles/barriers to replicated Google’s data infrastructure and amassing its dataset than honing the ML</p><p>Jason Black [3:09 PM] <br>Just as much as Google doesn’t directly make money on those algos, so too can other startups leverage the infrastructure so long as they don’t plan on making money on the commoditized portion</p><p>Morgan [3:10 PM] <br><a href="http://twitter.com/markjohnson">@markjohnson</a>: outsourcing TensorFlow</p><p>Alex [3:10 PM] <br>I think it’s a pretty safe move for Google to do something like this given their market share. I think Tesla’s open sourcing of their patents up against GM and big Auto is a much riskier bet.</p><p>Sandy Steier [3:10 PM] <br>You need (1) large amounts of data, (2) the ability to handle it (store, preprocess, etc.), (3) the right algorithms, (4) people who know what they are doing, and (5) an organization that can act on the results.</p><p>Jason Black [3:10 PM] <br>I think the more disruptive thing is Google opening up some part/all of its data streams</p><p>Mark Johnson [3:10 PM] <br>let’s be clear: there’s a big difference between open sourcing algorithms and open sourcing the ML features that you discover.</p><p>[3:11] <br>remember that Google’s original algos were published in a paper — they have a long history of doing that</p><p>[3:11] <br>but how you combine all those features is super tough — again, what % of searches are getting better and what features (and combination of features) are contributing to that</p><p>Sven [3:11 PM] <br>Another reason for open sourcing TensorFlow is that they can now recruit people that have prior exposure to TensorFlow. DeepMind and FAIR use Torch which is open source and you can have an intelligent discussion with a candidate about it because they have all used it before.</p><p>Jason Black [3:12 PM] <br>have any of the entrepreneurs in the room tinkered with/used TensorFlow or MSFT’s CNTK? (edited)</p><p>Sven [3:13 PM] <br>I have tinkered with TensorFlow.</p><p>Jason Black [3:14 PM] <br><a href="http://twitter.com/svenkreiss">@svenkreiss</a>: As something you’d leverage? Internally? Sounds like it might help at Descartes Labs as well (cc <a href="http://twitter.com/markjohnson">@markjohnson</a>)</p><p>Alex [3:14 PM] <br>are their numbers released on how widely used or implemented tensorflow has been post-open source?</p><p>Mark Johnson [3:15 PM] <br>we’ve played around with Caffe… good for prototyping but not sure we’d use it in production yet</p><p>Sven [3:16 PM] <br>I was looking into it in my free time. Wanted to know it better. I would feel more comfortable using TensorFlow in production than Torch, but right now we don’t have a use case for it. (edited)</p><p>Sandy Steier [3:17 PM] <br>I looked briefly at TensorFlow. The issue I see is that it is technical (like Hadoop), which brings up my point (4) above. If ML is to be really successful, it needs to be democratized. Someone would need to build a more approachable app on top of TensorFlow or its ilk.</p><p>Phil Boyer [3:18 PM] <br>Do you guys think that open source algorithms will have as big of an impact as open source infrastructure / DBs?</p><p>Mark Johnson [3:18 PM] <br>thinks there’s always a hot new algorithm and it can be a distraction from what you really want, which is results. It’s not the algorithm but as per <a href="http://twitter.com/sandysteier">@sandysteier</a>’s (4), people who know what to do with it.</p><p>Jason Black [3:18 PM] <br>Correct me if I’m wrong, but given the complexity of the task TensorFlow seems like a step in the right direction towards simplicity. Just from a documentation/tutorial standpoint, it’s pretty impressive what TF launched with</p><p>Sandy Steier [3:20 PM] <br>TensorFlow is the first step in what seems like a journey of a thousand miles.</p><p>Sven [3:20 PM] <br>The complexity of ML is another reason why open sourcing TF was smart: There is still a lot of research in Deep Learning. Even if you decide you want to put the effort in and build a product with it, you want to benchmark against a published reference implementation. Those are currently mostly in Torch.</p><p>Morgan [3:22 PM] <br>investors: how do you feel about investing in a startup whose ML implementation is dependent on TensorFlow?</p><p>Sven [3:24 PM] <br>versus Caffe or Torch or versus more traditional tools?</p><p>Phil Boyer [3:24 PM] <br>If a company utilizes TensorFlow, I personally wouldn’t think of it any differently than utilizing an open source database. If there are things you can leverage from the OSS, why recreate the wheel?</p><p>Morgan [3:24 PM] <br>versus writing algos in-house</p><p>[3:24] <br><a href="http://twitter.com/markjohnson">@markjohnson</a>: explain the thumbs down?</p><p>Sandy Steier [3:24 PM] <br>I’m not a VC but I agree with Phil — as long as the startup adds something valuable of its own.</p><p>Morgan [3:25 PM] <br><a href="http://twitter.com/svenkreiss">@svenkreiss</a>: tool agnostic, i’m mainly curious about startups whose core “machine learning tech” is calling an API…I’ve seen this mostly with IBM Watson</p><p>Mark Johnson [3:25 PM] <br>I’m nervous about just claiming to be a Deep Learning company while using standard infrastructure. Deep Learning promises to end the task of feature selection, and I think that’s still a core and incredibly important part of deep learning.</p><p>[3:27] <br>Deep Learning is a fad right now and I’m not down with fads… I’m down with what works. I remember when boosted trees were hot and we wasted months going down that path when a simpler algo performed way better. Beware companies with religion around a technique.</p><p>Jason Black [3:27 PM] <br>Depends on the startup and use case. I think if it is more a play at a space that will require top performance, building off of an open source and potentially more generic base decreases the flexibility you might have long term (since you don’t own the whole stack) and has the potential limit the your performance (in some cases). If you are leveraging it for a more vertically oriented business application that just happens to leverage ML to assist, I’d be more comfortable.</p><p>Josef Feldman [3:27 PM] <br><a href="http://twitter.com/morganpolotan">@morganpolotan</a>: Google has abandoned projects in the past so I’d think there’s a general danger of over dependance on an opensource community unless you actively manage that community (Nodesource/node.j and Automattic/wordpress being two good examples).</p><p>Shivon Zilis [3:28 PM] <br>Sorry for being late! Trying to catch up on all the brilliance :simple_smile:</p><p>Jason Black [3:29 PM] <br>Welcome, <a href="http://twitter.com/shivon">@shivon</a>!</p><p>David Wolfson [3:29 PM] <br>joined data by invitation from <a href="http://twitter.com/josef_feldman">@josef_feldman</a></p><p>Jason Black [3:30 PM] <br>Unfortunately, I’ve got to roll to a call. Please continue the discussion! Really liked the first session and would appreciate feedback. Feel free to DM me. Cheers!</p><p>Sandy Steier [3:30 PM] <br>Arguably opensource is protection against Google (or anyone else) abandoning a software direction. But I’m personally not a huge fan of opensource so I am not going to argue that too strongly.</p><p>Josef Feldman [3:32 PM] <br>Thanks <a href="http://twitter.com/jason">@jason</a> for leading this convo!</p><p>Alex [3:32 PM] <br>i’ve got to jump too, interesting stuff. thanks guys!</p><p>Morgan [3:33 PM] <br>thanks for organizing <a href="http://twitter.com/jason">@jason</a></p><p>[3:33] <br><a href="http://twitter.com/sandysteier">@sandysteier</a>: why are you not a huge fan of opensource?</p><p>Phil Boyer [3:36 PM] <br>heading out too guys. good stuff!</p><p>Josef Feldman [3:38 PM] <br><a href="http://twitter.com/sandysteier">@sandysteier</a>: here’s a study that was done on OSS projects, 152k dormant projects on source forge. <a href="http://staff.lero.ie/stol/files/2013/03/2013-Is-It-All-Lost-A-Study-of-Inactive-Open-Source-Projects.pdf">http://staff.lero.ie/stol/files/2013/03/2013-Is-It-All-Lost-A-Study-of-Inactive-Open-Source-Projects.pdf</a></p><p>Sandy Steier [3:38 PM] <br>It’s hard to maintain something that is an amalgam of efforts. It’s usually not the best stuff out there (if it were, it would be kept under wraps). It’s often a camel (you know, a horse built by a committee). Having said that, I think it’s fine for basic stuff (like Javascript widget libraries), but I wouldn’t use it for core functionality. It think the disappointment with Hadoop (which was obvious to me from the start) is a good example.</p><p>Morgan [3:41 PM] <br>good point</p><p>Sandy Steier [3:45 PM] <br>I’m going to hop…</p><p>Morgan [3:55 PM] <br>ciao <a href="http://twitter.com/sandysteier">@sandysteier</a>!</p><p>Heston Berkman [5:03 PM] <br>joined data by invitation from <a href="http://twitter.com/josef_feldman">@josef_feldman</a>, along with <a href="http://twitter.com/graham">@graham</a>, <a href="http://twitter.com/joelusv">@joelusv</a>, <a href="http://twitter.com/jayfarber">@jayfarber</a>, <a href="http://twitter.com/yanayasevich">@yanayasevich</a> and some others.. Also, <a href="http://twitter.com/ann">@ann</a> joined, <a href="http://twitter.com/valerygrx">@valerygrx</a> joined, <a href="http://twitter.com/lawrence">@lawrence</a> joined, <a href="http://twitter.com/peter">@peter</a> joined along with some others.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=911281facf36" width="1" height="1" alt=""><hr><p><a href="https://medium.com/venture-capital-think-tank/the-role-of-open-source-ml-platforms-in-the-startup-ecosystem-conversation-with-sandy-steier-911281facf36">The role of open source Machine Learning platforms in the startup ecosystem— Conversation with…</a> was originally published in <a href="https://medium.com/venture-capital-think-tank">ThinkTank.vc</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Chat with Mat Dellorso (CEO of Wealthforge) discussing Title III of the JOBS Act]]></title>
            <link>https://medium.com/venture-capital-think-tank/chat-with-mat-dellorso-ceo-of-wealthforge-discussing-title-iii-of-the-jobs-act-2c01ea0d6e7?source=rss-efa742544b48------2</link>
            <guid isPermaLink="false">https://medium.com/p/2c01ea0d6e7</guid>
            <category><![CDATA[wealthforge]]></category>
            <category><![CDATA[title-iii]]></category>
            <category><![CDATA[crowdfunding]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[jobs-act]]></category>
            <dc:creator><![CDATA[Josef Feldman]]></dc:creator>
            <pubDate>Tue, 17 Jan 2017 21:32:44 GMT</pubDate>
            <atom:updated>2017-01-17T21:32:44.648Z</atom:updated>
            <content:encoded><![CDATA[<p>Jay Farber [2:01 PM] <br>Hi all! Welcome to the next installment of the ThinkTank.vc FinTech chat. Today we’ll be discussing Title III of the JOBS Act, with Mat Dellorso (CEO of Wealthforge) here to join. If you haven’t read up, here’s a quick primer on the subject: <a href="http://fortune.com/2016/05/16/title-iii-jobs-act/">http://fortune.com/2016/05/16/title-iii-jobs-act/</a><br>Fortune Lucinda Shen<br>Now Anybody Can Try Being a Venture Capitalist<br>Barriers to entry come crashing down.</p><p>[2:02] <br><a href="http://twitter.com/mdellorso">@mdellorso</a>: mind sharing a few opening thoughts on what you’ve seen so far in the couple weeks since it went into place?</p><p>Mat Dellorso [2:02 PM] <br>joined fintech by invitation from <a href="http://twitter.com/jayfarber">@jayfarber</a></p><p>Mat Dellorso [2:04 PM] <br>Yes glad to thank you. So on May 16th after 4 long years of rule making the SEC promulgated Title III aka crowdfunding rules allowing small businesses and startups to raise up to $1M from “the crowd” allowing non-accredited investors the ability to invest in startups and private companies for the first time</p><p>[2:04] <br>is everyone aware of the rules and regulations or are their questions about that?</p><p>Jay Farber [2:05 PM] <br>I think we have the high-level view (though certainly in less detail than you). Maybe worth sharing what you’ve seen so far in terms of platforms starting to take advantage of it?</p><p>Mat Dellorso [2:06 PM] <br>Some thoughts that we have at WealthForge, in being active participants in online capital formation over the past 6 years are that the new regulations are restrictive and probably only advantageous to certain types of businesses such as consumer products companies, or those around an affinity or location so that the community would invest in it.</p><p>Josef Feldman [2:06 PM] <br>I’m curious what the general opinion is about the types of companies / projects that will most benefit from crowdfunding. People I’ve spoken to generally think there’ll be a severe adverse selection problem (esp for companies that can obtain traditional venture financing)</p><p>Mat Dellorso [2:08 PM] <br>I believe for the vast majority of offerings there may be an adverse selection problem as yes in many scenarios VC funding or a large investment strategic partner could be a better investment option for the best companies that can receive it. However, we have seen several scenarios where a company could benefit from many investors, to spread the word, such as in the previous case of Oculus or Pebble benefiting from highly publicized support from many backers or potential investors</p><p>[2:09] <br>The companies that likely utilize title III and have investor interest may not look like VC type investment opportunities and returns, likely the offering would be for a different purpose such as awareness, or like Sam Adams did many years ago to include their customers and supporters in a tranche of the offering</p><p>Josef Feldman [2:09 PM] <br>yes but in those cases they benefit from traditional crowdfunding not necessarily equity crowdfunding</p><p>Chad Castro [2:10 PM] <br>Forgive the 101 question, Mat but investors making &lt;$100,000 per year can invest the greater of $2,000 or 5% of annual income. Investors making &gt;$100,000 per year can invest up to 10% of their annual income — Who’s job to track/verify that — the registered intermediary?</p><p>Mat Dellorso [2:10 PM] <br>Yes equity crowdfunding is brand new, yet many of the backers of Oculus sure would have rather had equity</p><p>[2:10] <br>Chad great questions</p><p>[2:11] <br>Right now I do not believe there is a centralized body that aggregates that information across portals and offerings</p><p>[2:11] <br>It is very much a protection upon investors to guard against themselves</p><p>[2:12] <br>At WealthForge, we have not done any title III offerings as these questions have yet to be sorted out and have yet to have clear regulation on process and practicality</p><p>Jay Farber [2:12 PM] <br>So question to the other investors <a href="http://twitter.com/rzullo">@rzullo</a> <a href="http://twitter.com/awalsh">@awalsh</a> <a href="http://twitter.com/drew">@drew</a>.aldrich <a href="http://twitter.com/josef_feldman">@josef_feldman</a> <a href="http://twitter.com/chad">@chad</a> — would you be OK with one of your companies / prospective companies raising part of their round from the crowd? Any scenarios where you’d encourage it?</p><p>[2:13] <br><a href="http://twitter.com/nkrobles">@nkrobles</a>:</p><p>Mat Dellorso [2:13 PM] <br>I believe the broker-dealer or registered funding portal, where all title III offerings must take place on, would take on that responsibility of making sure the investors are suitabile ie have not invested more than the thresholds</p><p>Josef Feldman [2:13 PM] <br><a href="http://twitter.com/mdellorso">@mdellorso</a>: have you seen any clever twists on crowdfunding, such as limiting participation to engineers or other true value add investors who may otherwise not have any ‘skin in the game’</p><p>[2:14] <br><a href="http://twitter.com/jayfarber">@jayfarber</a>: I’d want to know how this looks on the cap table and if there are any long term regulatory requirements due to having retail investors involved</p><p>Drew Aldrich [2:14 PM] <br>Yes. I like equity as turning passionate, still somewhat affluent customers into true evangelists (apologize for being slow….have to type on cell).</p><p>[2:14] <br>So I like companies that are B2C, and “community based” i.e. Able for people to recruit others to the compmah</p><p>Mat Dellorso [2:15 PM] <br>Josef- I have heard of one clever offering that was discussing doing that in concept. I believe only 40 or so Form C’s — the form you need to file in advance of doing a title III raise — have been approved thus the use cases are just emerging</p><p>Drew Aldrich [2:15 PM] <br>Vcs to price the round, but carve out a small amount for existing customers</p><p>Jay Farber [2:16 PM] <br>One interesting angle could be secondaries. If I’m a VC I can either put $3M in now, or put $1M in with $2M from the crowd, and if I like it wait for the crowd to sell their shares to me as an effective ‘call option’ on the company</p><p>Mat Dellorso [2:16 PM] <br>One of the many limitations we see with the current regulations are that the investors, although exempt from the reporting number towards being a public company, all must be individually on the companies cap table so no SPV’s pooling into one entry on the cap table and how do you manage voting rights, distributions etc. Would most likely have to be a special class of stock with that governance</p><p>Josef Feldman [2:16 PM] <br><a href="http://twitter.com/mdellorso">@mdellorso</a>: are form C’s publicly listed? someone should create an aggregator for that!</p><p>Mat Dellorso [2:16 PM] <br>Yes for C data is publicly available via the SEC edgar database</p><p>[2:16] <br>we spend a lot of time analyzing Form D data at WealthForge</p><p>Chad Castro [2:17 PM] <br>Reg issues aside, won’t cap of $1M in a rolling 12-month period tamp enthusiasm for this type of funding?</p><p>Mat Dellorso [2:17 PM] <br>Also for A-1 is publically available for also relatively new Reg A+ data</p><p>[2:17] <br>Chad — yes we believe most companies over time will need to raise more. Issuers can do a Reg D offering as well</p><p>Jay Farber [2:18 PM] <br>Who here would consider actually making a Title III investment of their own $?</p><p>[2:19] <br>…bueller?</p><p>Mat Dellorso [2:19 PM] <br>I would, depends on the offering</p><p>Josef Feldman [2:19 PM] <br><a href="http://twitter.com/mdellorso">@mdellorso</a>: Have you heard of anyone trying to do a crowdfunded fund to invest in crowdfunding companies? an idea I had years ago when title II was starting to be discussed.</p><p>[2:20] <br>I think it could work</p><p>Mat Dellorso [2:20 PM] <br>So the fund would have to be done through Reg D or Reg A and then could invest in Reg CF offerings</p><p>Rick Zullo [2:20 PM] <br>Wasn’t alpha (Nick Chirls old shop) trying to replicate that passionate group of consumers around crwodfunding a product?</p><p>Drew Aldrich [2:20 PM] <br>I looked at the the gimlet offering…but passed</p><p>Rick Zullo [2:21 PM] <br>For me, that’s a lot more interesting than B2B (which is 99% of what I do) and I usually take it as a bad sign if someone has to crowdfund</p><p>Mat Dellorso [2:21 PM] <br>I have heard of and had ideas for a similar concept Josef, for broad diversification just Title III offerings are still new and relatively few thus far. Then there is the concept of direct investing vs the fees and management of a fund</p><p>Jay Farber [2:21 PM] <br><a href="http://twitter.com/drew">@drew</a>.aldrich: Looking like a decent bet so far</p><p>Mat Dellorso [2:21 PM] <br>what these new regs do is allow for direct investing which is a unique phenomenon, yet the advice and transparency to sort through the offerings is not there yet = might still need analysts</p><p>Jay Farber [2:22 PM] <br>Interesting…who is the right equivalent of equity researchers in the public markets?</p><p>[2:22] <br>Things move so much more quickly, are illiquid, and much less data</p><p>[2:22] <br>And fewer restrictions around pumping a stock you own (or a deal you’re doing), I’d assume</p><p>Mat Dellorso [2:22 PM] <br>yes and on small offerings who pays for research?</p><p>Jay Farber [2:23 PM] <br>Maybe people doing the deal just publish the diligence work they did</p><p>[2:23] <br>Happens all the time when you get pulled in as a VC to a well-baked deal</p><p>Josef Feldman [2:23 PM] <br>I think angellist has accomplished this with their syndicate model as someone needs to lead diligence.. opening this up to the masses when no one particular party has enough economic incentive to truly dig in sounds a bit risky..</p><p>Mat Dellorso [2:24 PM] <br>Yes a lead or priced deal could make sense, a deal without a lead investor or no backing seems to be hard</p><p>Nikko Karlo Robles [2:24 PM] <br><a href="http://twitter.com/mdellorso">@mdellorso</a>: Just curious, any thoughts what “scale” looks like when Title III is in order? Like how many non-accredited investors are do crowd-equity platforms look to engage with?</p><p>Mat Dellorso [2:25 PM] <br>Real estate is another interesting vertical for crowdfunding. If a sponsor allows a community to invest and earn a projected IRR of say 8% thats not bad for a non accredited investor who cannot buy or invest in real estate with small dollars</p><p>[2:26] <br><a href="http://twitter.com/nkrobles">@nkrobles</a>: good question. In our Reg D offerings we often see 30 investors per million. In CF offerings I would imagine that number would be multiplied by 10 or 100x</p><p>Rick Zullo [2:26 PM] <br>but how can you market a 8% IRR security</p><p>[2:26] <br>with out significant protections around it</p><p>[2:26] <br>lack of knowledge around default rates, etc.</p><p>Jay Farber [2:27 PM] <br>Probably listed as 8% coupon, so to speak. Not risk-adjusted.</p><p>[2:27] <br>That’s how I’ve seen RE deals at least</p><p>[2:28] <br>Alright — let’s be creative in these last few minutes. What’s your best pitch for a crowdfunding platform you think could effectively leverage title III? Here’s mine: collective diligence. You work as an online angel group where you each come from an industry (say, retail), and each ask your company if they would stock a given consumer product. Whoever wants in on the deal gives an answer from their company, and the deal only goes through if enough people’s answers are positive</p><p>Josef Feldman [2:29 PM] <br><a href="http://twitter.com/rzullo">@rzullo</a>: I think this is also a reflection of the adverse selection problem as well. The general trend with P2P has always been to move upstream and towards institutional capital. Why would a good deal be listed on a crowdfunding site? The crowds do not seem to have as much access to capital or diligence as institutional investors who exist to provide this function (consumer products / Kickstarter type campaigns being the exception)</p><p>Mat Dellorso [2:30 PM] <br>Also with regard to scale question. About 40,000 companies raised money via Reg D in 2015. If everyone of those companies chose to do a Reg CF, thats only 40B at scale vs Reg D which is a $1T market annual</p><p>Rick Zullo [2:30 PM] <br>Isnt that how ethereum was funded?</p><p>[2:30] <br>thought I read an article about that the other day, 100% anonymous vote</p><p>[2:31] <br>For me, I think pooled portfolios and later stage companies make a lot more sense</p><p>Mat Dellorso [2:31 PM] <br>I do think there are angel group 2.0 type applications</p><p>Rick Zullo [2:31 PM] <br>I also wonder when VC funds will be able to crowdsource capital</p><p>Gul [2:31 PM] <br><a href="http://twitter.com/jayfarber">@jayfarber</a>: I’d be super interested in joining an investor syndicate w/ a small amount contribution, especially in a space that I understand well. As things take off, I think whats personally interesting to me is how this enables you to tap your networks for investment opportunities in promising startups (even if your network is not inv. or venture focused)</p><p>Rick Zullo [2:31 PM] <br>effectively already happening with angelist syndicate</p><p>[2:32] <br>ditto to gul</p><p>Mat Dellorso [2:32 PM] <br>SPV’s are capped at 99 investors</p><p>Gul [2:33 PM] <br><a href="http://twitter.com/jayfarber">@jayfarber</a>: thanks for organizing this J. Super interesting / informative. Looking forward to a follow up chat on inv. opportunities w/ promising Fintech startups!</p><p>Josef Feldman [2:33 PM] <br><a href="http://twitter.com/jayfarber">@jayfarber</a>: fun! I’d say a platform that copies what Juno is trying to do to Uber but across all sorts of industries.. that’d be a fun experiment</p><p>Jay Farber [2:33 PM] <br>Hey all, we can keep chatting here but for those who have to drop off now — thanks for joining! Stay tuned for part 3 coming soon…</p><p>Josef Feldman [2:34 PM] <br>Thanks <a href="http://twitter.com/jayfarber">@jayfarber</a> for moderating and <a href="http://twitter.com/mdellorso">@mdellorso</a> for leading. this was great</p><p>[2:36] <br><a href="http://twitter.com/channel">@channel</a>: taking suggestions for next topic to cover if anyone has requests / particular interests within Fintech</p><p>Mat Dellorso [2:36 PM] <br>Thank you Josef and Jay</p><p>Rick Zullo [2:38 PM] <br>thanks guys!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2c01ea0d6e7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/venture-capital-think-tank/chat-with-mat-dellorso-ceo-of-wealthforge-discussing-title-iii-of-the-jobs-act-2c01ea0d6e7">Chat with Mat Dellorso (CEO of Wealthforge) discussing Title III of the JOBS Act</a> was originally published in <a href="https://medium.com/venture-capital-think-tank">ThinkTank.vc</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Marketplaces Chat w/ Founders of OLX (Fabrice Grinda), Hired (Allan Grant), Inturn (Ronen Lazar)]]></title>
            <link>https://medium.com/venture-capital-think-tank/1-marketplaces-chat-w-founders-of-olx-fabrice-grinda-hired-allan-grant-inturn-ronen-df1bfbc95467?source=rss-efa742544b48------2</link>
            <guid isPermaLink="false">https://medium.com/p/df1bfbc95467</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[hired]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[marketplaces]]></category>
            <category><![CDATA[founder-stories]]></category>
            <dc:creator><![CDATA[Josef Feldman]]></dc:creator>
            <pubDate>Mon, 02 Jan 2017 20:04:28 GMT</pubDate>
            <atom:updated>2017-01-17T21:34:49.321Z</atom:updated>
            <content:encoded><![CDATA[<p>Neeraj Singhal [12:09 PM] <br><a href="http://twitter.com/channel">@channel</a> — A little over 20 minutes before we kick-off our marketplaces chat with Fabrice Grinda, Allan Grant, and Ronen Lazar.</p><p>As a quick reminder, here are the question prompts for the discussion:<br> — What are the right benchmarks for assessing liquidity in your marketplace?<br> — How has your business adopted strategies for achieving supply, in markets where it is not sticky or persistent? How have you customized this process for different geographies?<br> — From a venture perspective, how do you model and value a company’s ability to acquire supply and demand?<br> — Which companies / trends are you most excited about?</p><p>Recent article on strategies to achieve liquidity: <a href="http://firstround.com/review/How-Modern-Marketplaces-Like-Uber-Airbnb-Build-Trust-to-Hit-Liquidity/">http://firstround.com/review/How-Modern-Marketplaces-Like-Uber-Airbnb-Build-Trust-to-Hit-Liquidity/</a></p><p>Prashant Fonseka [12:25 PM] <br>hi everyone</p><p>Vinay Trivedi [12:27 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a></p><p>Neeraj Singhal [12:30 PM] <br>Alright guys, I think we can get started… thanks so much for taking the time to join everyone</p><p>[12:31] <br>I’ll quickly introduce our panelists: <a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> — Founder @ FJ Labs, Fmr. CEO @ OLX<br><a href="http://twitter.com/allan">@allan</a> — Founder @ Hired<br><a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> — Founder @ Inturn</p><p>Josef Feldman [12:31 PM] <br>Thanks Neeraj! Will share a list of all our participants shortly as well. Hello all</p><p>Allan Grant [12:32 PM] <br>Hi guys. Thanks for having us!</p><p>Neeraj Singhal [12:32 PM] <br>Idea is pretty simple here… tee up a good topic with relevant folks and have an interactive discussion</p><p>Ronen Lazar [12:32 PM] <br>Hi! Thanks for hosting this panel</p><p>Neeraj Singhal [12:32 PM] <br>With that in mind… let’s kick off… I’ll start with you Allan, what are you looking at in terms of benchmarks for liquidity in your business (Hired.com)</p><p>Allan Grant [12:33 PM] <br>For us at Hired, we have a singular mission of “getting people jobs they love” — so the ultimate sign of liquidity is people actually getting hired.</p><p>[12:34] <br>That’s a late funnel metric and there are some early indicators that are important to getting there</p><p>[12:35] <br>On Hired, companies send “interview requests” to candidate, which starts the whole process of generating a hire. So one way to tell how much liquidity exists in a market is based on the number of interview requests (IVRs) in a given time period.</p><p>Stephen Yang [12:35 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a>, along with <a href="http://twitter.com/davewolfson">@davewolfson</a></p><p>Allan Grant [12:35 PM] <br>Because talent can be segmented into different roles, skill sets, and locations — we don’t think of ourselves as a single marketplace; we are a collection of many markets.</p><p>[12:36] <br>So we measure liquidity across each sub-market — whether early signs such as IVRs, ultimate results such as hired, or intermediate metrics between those two.</p><p>[12:36] <br>I.e. Ruby engineers in SF, or designers in NYC.</p><p>[12:36] <br>One last thing to add…</p><p>[12:37] <br>In addition to measuring liquidity — which are ultimately “transaction metrics” — we also measure the two sides: supply and demand. Those are measured independently and probably a topic for another discussion.</p><p>Neeraj Singhal [12:37 PM] <br><a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> &amp; <a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> feel free to chime in on your company and the portfolio of companies you’ve built.</p><p>Josef Feldman [12:37 PM] <br><a href="http://twitter.com/allan">@allan</a> sounds tricky to master! is there anything you guys do on a geographic basis to monitor / balance supply &amp; demand?</p><p>Evan Maclin [12:38 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a>, along with <a href="http://twitter.com/marc">@marc</a>. Also, <a href="http://twitter.com/nkrobles">@nkrobles</a> joined.</p><p>Allan Grant [12:38 PM] <br>We track engagement of each “unit” (candidate or employer) individually to see if they are getting what they want.</p><p>[12:39] <br>For example is the candidate being viewed and getting IVRs? Is the employer browsing and reaching out to candidates.</p><p>Deepak Masand [12:39 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a></p><p>Allan Grant [12:39 PM] <br>A market may have liquidity for most participants but still be missing out on needs of a few players.</p><p>[12:40] <br>If we can’t generate the liquidity then we don’t let those players into the market until we can — that’s our “curation” approach.</p><p>[12:40] <br>Better to not let someone in (yet) than to disappoint them with the experience.</p><p>Fabrice Grinda [12:41 PM] <br>I will answer separately for OLX and companies we are evaluating from an investment perspective. In both cases ultimately, like Allan we look at late funnel metrics and notably the sell through rate, then it’s a question of scale.</p><p>For OLX we shoot for a 40% sell through rate in the general goods category at scale. At scale on average means 5 net new lister / 1,000 capita / day and a 75% marketshare in the country. We end up being in the top 5–10 sites in the countries where we accomplish this. We track this liquidity city by city and category by category. It typically means we end up with 15–30% of the population of the country using the site every month.</p><p>From an investment perspective when we invest in 2 sided marketplace at an early stage — say $100–500k / month in GMV if they have a 25% sell through rate then it’s a good sign of early liquidity. Of course depending on what they are selling it maybe at a hyperlocal or regional or national level.</p><p>Jay Farber [12:41 PM] <br>So <a href="http://twitter.com/allan">@allan</a> , when you enter a new ‘market’, how do you seed both sides of the market to go from zero to 100?</p><p>Ronen Lazar [12:41 PM] <br>So INTURN is a B2B marketplace which addresses the complexities of the off price space with initial focus on apparel and footwear. Our marketplace is inherently atypical due to the quantity, depth and quantity of inventory packaged together, confidentiality of the sales efforts to protect brand integrity and the steep decline in inventory value over a short period of time. What’s even more astounding is that price is not always the determining factor for a sale.</p><p>Victor Wang [12:42 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/benjaminzeitz">@benjaminzeitz</a></p><p>Fabrice Grinda [12:43 PM] <br>The way we seed varies from marketplace to marketplace. For most we start by getting the supply or sellers because they have a financial incentive to be on the platform and are ok with us not having demand yet. Depending on the vertical we do it through a combination of TV ads, online ads or even a sales team.</p><p>[12:43] <br>Whatever the economics support…</p><p>Ronen Lazar [12:43 PM] <br>Our sell through rates are unusually high as the commodity is product that was selling at a specific price just days earlier and now discounted anywhere from 40–80% off.</p><p>Fabrice Grinda [12:43 PM] <br>at a unit economic level</p><p>Allan Grant [12:45 PM] <br><a href="http://twitter.com/jayfarber">@jayfarber</a>: we look at entering a new market and achieving initial liquidity as a different “playbook” from growing a market. Specifically our playbook entails first focusing on queuing up demand (companies) since their hiring needs won’t go away while we built up supply, then we build up matching supply, and then we “pre-launch” the market by driving both sides to show up on the same day. Our cadence is new groups of candidates starting every week, so the specific start time makes sense.</p><p>[12:46] <br>The market is then in soft launch for 3–4 months as we gradually increase supply for the demand we have (since a company can hire many people, but a candidate only gets one job), as well as bringing in other facets of demand (companies hiring different skill sets).</p><p>[12:47] <br>Once we hit a target goal of liquidity, we announce that the market is now launched, and do PR around it — driving more on both sides knowing we won’t disappoint.</p><p>Neeraj Singhal [12:47 PM] <br>Very interesting… at Uber, we have very specific liquidity goals in mind for operating our cities and countries. Different model of marketplaces, and perhaps easier given it’s a very clear 2-sided marketplace, but we looked at Supply liquidity, which is essentially, are there enough cars on the road for a rider to get a car in under 5 minutes (this varied by country/city… for example, India’s benchmark was 8 minutes, whereas US cities was 3 minutes). and on the demand side, we had to make sure the total amount of fares accumulated for a driver was greater than the next alternative opportunity to earn, making them just a little over ‘indifferent’ between choosing Uber as a platform to earn and another hourly earning opportunity.</p><p>Jay Farber [12:47 PM] <br>Interesting…don’t expose the two sides to each other (launch) until the sell through rate would hit a minimum threshold</p><p>Allan Grant [12:48 PM] <br>Exactly</p><p>Neeraj Singhal [12:49 PM] <br><a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> — How have you gone about boosting supply in new markets, and what are your early thoughts on ways to boost this?</p><p>[12:51] <br>And <a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> — While Ronen is typing, since you also represent the Venture side of our panel… how do you model and value a company’s ability to acquire supply and demand?</p><p>Ronen Lazar [12:51 PM] <br>the most complex part of our marketplace is the lack of consistent process on both the supply and demand sides</p><p>James Holcome [12:51 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/nick">@nick</a></p><p>Ronen Lazar [12:51 PM] <br>So we’ve created efficiency layers for both pre and post transaction</p><p>[12:53] <br>On the supplier side (brands) our platform becomes their visibility into their business and allows them to be more proactive. We enable them to curate assortments of goods and include all the appropriate data and visuals. The platform allows them to analyze margins continuously and once a deal is done, enter orders. All this was manual prior to us launching.</p><p>[12:54] <br>So the suppliers are excited even to benefit on the efficiency we offer as it creates both tangible and intangible benefits instantly</p><p>[12:55] <br>So we’ve taken a very systematic approach to build market networks surrounding suppliers that are private or pubic and help them innovate their processes</p><p>Fabrice Grinda [12:55 PM] <br>We are very unit economic driven. We want the company’s net contribution margin per customer over the first 18 months to be 3–4x the fully loaded customer acquisition cost of supply + demand. Let’s say they use Google as a successful paid channel where the economics seem to work based on the early cohorts we are going to check that there is enough volume in the channel that they can keep scaling. We have come to realize most businesses don’t have viral coefficients above 1 so paid acquisition is key and it’s important to find both channels where the economics work and that are scalable.</p><p>A bit of a non-sequitur, but one of the ways we found to dramatically lower CACs is to offer a super verticalized experience and by really talking the talk of the suppliers the way Mindbody has done it in a few verticals.</p><p>[12:56] <br>From a unit economic perspective btw it typically means for the marketplace to work you need either a high average order value or recurrence (or both).</p><p>[12:56] <br>That’s why I am skeptical of models like Shyp</p><p>Neeraj Singhal [12:57 PM] <br><a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> — Has this been a tough process in your International markets?</p><p>Josef Feldman [12:57 PM] <br><a href="http://twitter.com/channel">@channel</a> Getting requests for the full attendee list. Here it is! Feel free to connect with each other directly.</p><p>Also we have one more addition to the panel — <a href="http://twitter.com/dkalt123">@dkalt123</a> founder of Reverb.com who will share his thoughts as well.</p><p>Fabrice Grinda, Co-founder, FJ Labs, Co-CEO, OLX<br>Allan Grant, Co-founder, Hired<br>Ronen Lazar, Co-founder and CEO, INTURN<br>David Kalt, Founder and CEO, Reverb.com<br>Alex Levin, VP of Expansion, Handy<br>Vinay Trivedi, Sandbox Network<br>Xing Xin, Business Development and Growth, Beepi<br>Tyler Infelise, Head of Product, Beepi<br>Luke Fryer, Founder, Harri.com<br>Travis Ing, Associate, Hearst Ventures<br>Gunnar Froh, Founder, Wunder<br>Li Jin, Investment Partner, Andreessen Horowitz<br>Guimar Vaca Sittic, Partner, FJ Labs<br>Tessa Wanders, Associate, FJ Labs<br>Arne Halleraker, FJ Labs<br>Nick Greenfield, COO, Paribus<br>Stephen Varady, Founder and CEO, Slang<br>Karthik Sridharan, Co-founder, Kinnek<br>Ilias Beshimov, VP of Growth, Good.Co<br>Sage Ramadge, Executive Director, InSITE Fellows<br>Dana Goldstein, Head of FP&amp;A, Etsy<br>Thomas Bailey, Founder, Hatch Capital<br>Matt Heiman, Investor, Greylock Partners<br>Joshua Goldstein, Co-founder, Underdog.IO<br>Evan Maclin, Business Development Manager, Zeel<br>James Holcombe, Expansion Manager, Zeel<br>Nick Ober, Growth, Zeel<br>Muzzammil Zaveri, Partner, KPCB<br>Jay Farber, Associate, F-Prime Capital Partners<br>Morgan Polotan, Investor at Bloomberg Beta<br>Mohamed Haouache, OuiOpen<br>Prashant Fonseka, Associate, Crunchfund<br>Neeraj Singhal, Entrepreneur-In-Residence, FoundersGuild<br>David Wolfson, Partner, FoundersGuild<br>Josef Feldman, Partner, FoundersGuild</p><p>Mohamed Haouache [12:58 PM] <br>Mohamed Haouache, Storefront</p><p>Ronen Lazar [12:58 PM] <br>Interesting it hasn’t been as difficult as we imagined. We launched our North American marketplace in May 2015 and by September we were operating in Europe.</p><p>[12:59] <br>On the supply side there are brands always looking to sell their goods outside their primary market. On the demand side, since off price isnt as heavily built up outside the US and most brands don’t produce product for the off price market internationally, there is a built in demand</p><p>Neeraj Singhal [1:00 PM] <br>Welcome <a href="http://twitter.com/davidmort">@davidmort</a>! Would love to hear your thoughts on assessing the right benchmarks for liquidity in your business, and how you’ve gone about building up your supply side.</p><p>[1:00] <br>Apologies. Welcome <a href="http://twitter.com/davidmort">@davidmort</a> !</p><p>David Kalt [1:00 PM] <br>Thanks Josef, Reverb is a marketplace where musicians can discover, and purchase instruments and gear to help them make their music.. A potential global audience of 100 million indviduals, representing approx. $25 million in global GMV a year. We approach our liquidity problem by thinking about the bid/ask spread in pricing.</p><p>[1:01] <br>We have between 500,000 and 750,000 listings on our site at any given moment and our challenge is pricing differentiation and discovery.</p><p>Neeraj Singhal [1:02 PM] <br>Welcome <a href="http://twitter.com/dkalt123">@dkalt123</a> — The other David is also welcome, but his response is not as urgent.</p><p>[1:02] <br>3rd times a charm.</p><p>David Kalt [1:02 PM] <br>Since there are thousands of similar listed guitars with different condition, age, finish, shipping cost, etc — — the ability to distinguish pricing transparency has been the greatest driver in our growth trajectory.</p><p>[1:03] <br>We have built an extensive price guide, and delivering tons of related content to help drive price transparency. we combine that with negotiation and messaging capabilities that empower buyers to get fair and accurate pricing with every order.</p><p>Neeraj Singhal [1:04 PM] <br><a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> <a href="http://twitter.com/allan">@allan</a> <a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> <a href="http://twitter.com/dkalt123">@dkalt123</a> — What are some interesting innovations you’re seeing, and what are you most excited about?</p><p>David Kalt [1:04 PM] <br>The more pricing transparency, the more musicians are willing to purchase their next piece of gear knowing that there is a market for it. Liquidity drives confidence which drives transactions.</p><p>David Mort [1:06 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/neeraj">@neeraj</a></p><p>Ronen Lazar [1:06 PM] <br>I’m most excited about data transformation innovations that remove integration requirements. In retail that is by far the biggest hurdle tech companies are facing. By removing this barrier and even better, enabling both sides of the marketplace to resonate with the offering based on their own language, it delivers immediate success.</p><p>David Kalt [1:06 PM] <br>Personalization Engines is a crucial technology that we are investing in to deliver the most relevance to our end-users.</p><p>Fabrice Grinda [1:07 PM] <br>One clear trend we are seeing is that the best practices of consumer facing marketplaces are now being brought to B2B marketplaces where you can have the holy trinity of high margins, high average order value and high recurrency. The key is picking the verticals where supply and demand are sufficiently fragmented (to avoid margin compression and disintermediation) and where the sales cycles are not too long. My favorite example of this is Flexport</p><p>[1:08] <br>But Zesty, Grubmarket, Partmyride all fall in this category</p><p>Neeraj Singhal [1:08 PM] <br><a href="http://twitter.com/dkalt123">@dkalt123</a> — Can you explain the concept of personalization engines in a bit more detail? Would love to hear more about how this can be applied to other marketplaces as well.</p><p>David Kalt [1:08 PM] <br>Old word marketing would focus on PERSONA’s and marketing strategy to groups of like minded individuals.</p><p>[1:09] <br>Our Personalization Engine tracks behavior and finds correlations, that can then create a customized browse/feed experience similar to Netflix.</p><p>Allan Grant [1:09 PM] <br>Within marketplaces, one thing that’s been interesting to see is the birth of new models which only need to build up one side, while tapping into an almost limitless pool on the other side — through paid advertising or hyper targeting enabled by large scale data analysis. For example in recruiting we have a competitor that analyzes publicly available code to identify the right candidates for their customers (companies hiring engineers) without having those candidates ever need to sign up. So they aren’t limited by needing to build up liquidity in each geographic and skill market one at a time — they can service clients practically all over the world. I’ve also seen early signs of similar models being worked on in online dating.</p><p>David Kalt [1:10 PM] <br>When done right, the user doesn’t know it or feel it, and is blown away by the experience.</p><p>Josef Feldman [1:10 PM] <br><a href="http://twitter.com/allan">@allan</a> brilliant. would love to see more examples</p><p>Fabrice Grinda [1:12 PM] <br>Other fun trends are the use of AI. One of our companies now uses image recognition to suggest the title, description etc. so you no longer need to type it in when placing an item on sale</p><p>Ronen Lazar [1:12 PM] <br><a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> Do you find that in some verticals where supply and demand is fragmented that there isn’t enough stickiness to keep people engaged and instead they continuously shop deals around?</p><p>Neeraj Singhal [1:12 PM] <br><a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> — Are there any undisrupted yet highly fragmented industries left? If there is, feel free to DM me the other folks here don’t see… :wink:</p><p>David Kalt [1:13 PM] <br>Agreed, using machine learning and AI in our personalization engine is essential.</p><p>Fabrice Grinda [1:13 PM] <br>Many more than you think. Shipping was totally ignored until Flexport (and Cargomatic and others came along)</p><p>[1:13] <br>And AI can be used for quality control / content filtering and much much more</p><p>Allan Grant [1:13 PM] <br>Another interesting thing is what I’ve been calling “2nd order network effects” (perhaps someone knows the correct term for what I’m referring to?). What I mean is moats that emerge through the accumulation of a large number of transactions over time. For example we have tracked the outcome of every single interview that any candidate from Hired has gone on, since 2012. We are building towards a critical mass that will enable us to correlate interview performance with third party data, such as skill assessments, in a way that competitors won’t be able to until they have gone through years of liquidity and very many transactions.</p><p>Fabrice Grinda [1:14 PM] <br><a href="http://twitter.com/allan">@allan</a> Totally agree. In addition to liquidity being a barrier to entry you can do a better job because you have more data</p><p>[1:14] <br>Another trend I love is marketplaces that offer tools that make people better at their job</p><p>[1:15] <br><a href="http://www.rev.com/">www.rev.com</a> is an example of that</p><p>Josef Feldman [1:15 PM] <br><a href="http://twitter.com/channel">@channel</a> if anyone has any questions for the panel feel free to chime in</p><p>Fabrice Grinda [1:15 PM] <br>The people use their app to do a better job at transciption</p><p>[1:15] <br>They become so much more productive</p><p>[1:15] <br>that they have no incentive to disintermediate or go to a competitor</p><p>Neeraj Singhal [1:16 PM] <br>Yup, if our attendees have any questions, we can use the last couple minutes here to tackle them. Free to free to shoot them out to the group.<br> Josef Feldman<br><a href="http://twitter.com/channel">@channel</a> if anyone has any questions for the panel feel free to chime in<br>Posted in #marketplacesNov 1st, 2016 at 1:15 PM</p><p>Fabrice Grinda [1:18 PM] <br>BTW Verticalization is continuing unabated. You have a mega trend for instance of more work being gig focused which you have companies like ShiftGig doing an amazing job with. But then for more specialized work we are seeing vertical gig marketplaces like Nomad Health for doctors</p><p>[1:19] <br>And the verticalization of Craigslist &amp; eBay has more legs to run than I anticipated. Reverb is a great example of that. Alongside Chrono24 and many others.</p><p>Allan Grant [1:20 PM] <br>I have a question I’d love to ask the other panelists — something I’ve been thinking a lot about over the past few years: what leads marketplaces to become monopolistic vs oligopolistic?</p><p>Sandy Steier [1:21 PM] <br>joined marketplaces by invitation from <a href="http://twitter.com/josef_feldman">@josef_feldman</a></p><p>Neeraj Singhal [1:21 PM] <br>Before we hit the hour mark here… Just wanted to thank our panelists <a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> <a href="http://twitter.com/allan">@allan</a> <a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> &amp; <a href="http://twitter.com/dkalt123">@dkalt123</a> — Sincerely appreciate you taking the time to give the group some insight and your perspective. And to the broader group, feel free to continue to use this channel to connect with folks and share articles/questions/thoughts. The idea behind ThinkTank is to connect people who are solving similar problems in varying businesses/industries.</p><p>[1:22] <br>People feel free to continue the discussion folks…</p><p>Ronen Lazar [1:22 PM] <br>Thanks for hosting Neeraj</p><p>Allan Grant [1:22 PM] <br>Thank you Neeraj, Fabrice, Ronen, David, and everyone attending! This has been a fun discussion.</p><p>Fabrice Grinda [1:22 PM] <br>Thanks!</p><p>Neeraj Singhal [1:23 PM] <br>My pleasure… and keen to hear responses to <a href="http://twitter.com/allan">@allan</a> ‘s question!</p><p>David Kalt [1:23 PM] <br>Thanks, great idea generation.</p><p>Fabrice Grinda [1:24 PM] <br><a href="http://twitter.com/allan">@allan</a> happy to discuss the monopoly question too. It’s been on my mind as well in terms of what it takes in terms of network effects (ever more buyers brings ever more sellers etc.), lower CAC from having existing scale / brand + secondary effects of having more data to provide a better user experience.</p><p>[1:24] <br>+ one you are larger and have scale you get more access to capital and you can continue attacking</p><p>[1:25] <br>so even if you were headed towards a 60/40 split your capital advantage keeps allowing you to end up with a near monopoly</p><p>[1:25] <br>BTW it was not obvious ex-ante that classifieds would be a near monopoly</p><p>[1:25] <br>It’s not like auctions you can post on multiple sites</p><p>[1:26] <br>But people don’t bother once a site works better. And because you can’t monetize in a 60/40 split, you really need 85%+ market share you just keep going until you win (or merge until you win)</p><p>[1:26] <br>We spend $500 milion in TV ads in Brazil between us and our competitors and then merged and it was 1+1 = 10</p><p>[1:27] <br>But then in classifieds you end up with a business with 50–75% EBITDA margin</p><p>[1:27] <br>The thing is you can go from losing $100 million / year to $100 million in EBIDTA in less than 18 months once you win</p><p>[1:27] <br>but that dynamic leads to a Nash equilibrium where everyone spends</p><p>[1:27] <br>If I spend you and you don’t I win</p><p>[1:27] <br>If you spend and I don’t I lose</p><p>[1:28] <br>So the dominant strategy is spending</p><p>[1:28] <br>so you end up with a game theoretical framework where everyone spends money like crazy</p><p>[1:28] <br>But the rewards are worth it</p><p>[1:28] <br>but to get OLX to where it is today it cost over $1 billion in TV ads</p><p>[1:28] <br>And LetGo is spending a lot as well in the US to break in vs. Craigslist</p><p>Allan Grant [1:30 PM] <br>I’ve come to a similar thesis. Primarily thinking about 3 factors: 1) liquidity bleed over, 2) lock-in on one side, and 3) cost of access to capital. Limited liquidity bleed over is what makes online dating oligopolistic — you need liquidity in different cities, age groups, cultures; whereas something like eBay has high bleedover between categories and geos. Good example of lock-in on one side is Airbnb’s instant book feature — a seller that turns it on gets more business but can’t cross list on VRBO. And access to capital let’s you purchase liquidity in other markets/subsidizing subliquid areas to provide a better supply/demand aggregation experience for both sides. (edited)</p><p>Fabrice Grinda [1:31 PM] <br>The thing is things like dating or classifieds could be monopolies at a city level</p><p>[1:31] <br>But once you win a city you just keep going</p><p>[1:31] <br>Even though having liquidity in NY does not help you in SF</p><p>[1:31] <br>for the most part</p><p>Allan Grant [1:31 PM] <br>That’s what I mean by oligopolistic.</p><p>Fabrice Grinda [1:31 PM] <br>But if you have more access to capital + expertise you go</p><p>Josef Feldman [1:31 PM] <br><a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> fanduel V draftkings is another good example</p><p>Fabrice Grinda [1:31 PM] <br>Yes it’s oligopolistic at the beginning</p><p>[1:32] <br>But I suspect it ends as a monoploy</p><p>[1:32] <br>as people merge + one gets lower cost of capital because of scale, quality of team etc</p><p>Nick Greenfield [1:32 PM] <br>Uber has a massive cost of capital advantage over most players in the on demand (food, transportation, delivery) space and can therefore outspend all competition in multiple verticals in every geography. Is there any way for smaller players to overcome that advantage?</p><p>Fabrice Grinda [1:34 PM] <br><a href="http://twitter.com/regularnick">@regularnick</a> By providing a better experience in secondary verticals for sure</p><p>[1:34] <br>What do yo uthink the revenue breakdown is at Uber on Eats vs transportation</p><p>[1:34] <br>I suspect it’s the B team on Uber Eats and a much loower priority</p><p>Nick Greenfield [1:35 PM] <br>100:1</p><p>Fabrice Grinda [1:35 PM] <br>So Postmates / Seamless etc. are probably ok especially since they can differentiate the offering</p><p>[1:35] <br>and quality of experience</p><p>[1:35] <br>etc.</p><p>[1:36] <br>But I would NOT want to be Lyft</p><p>[1:36] <br>or Gett</p><p>[1:37] <br>Especially since Uber could use profitable markets to subsidize supply and/or demand in other markets. Lyft exists at the leisure of Uber. Now Uber may want them around for regulatory reasons</p><p>[1:37] <br>Airbnb is probably a global monopoly btw because people travel internationally.</p><p>[1:37] <br>So liquidity is cross border and cross country</p><p>[1:37] <br>cross city</p><p>[1:37] <br>Uber is more city by city</p><p>[1:38] <br>But if you win NY you can use profits there to subsidize a Lyft city for instance</p><p>[1:38] <br>That’s why I think the oligopoly setup is temporary</p><p>[1:38] <br>and it ends up essentially as a monopoly</p><p>Nick Greenfield [1:38 PM] <br>Isn’t this all true to the point of EBIDTA mattering?</p><p>[1:39] <br>once you have to grow EBIDTA, isn’t it very challenging to continue attempted monopolistic practices to thwart upstart local competition with pricing and subsidies</p><p>Fabrice Grinda [1:39 PM] <br>Well let me give you another example</p><p>[1:40] <br>OLX is now super profitable in 8 countries with EBITDA margin of over 75% (and I mean hundreds of millions of EBITDA / year)</p><p>[1:40] <br>It makes it reasonably easy to then invest tons of money in the next batch of countries we want to win</p><p>[1:40] <br>Like India</p><p>[1:40] <br>Or the US with LetGo</p><p>[1:41] <br>Sorry 50–75% EBITDA margin. Hundreds of millions of EBITDA</p><p>[1:41] <br>So you can still grow EBITDA and invest (or if your investors are patient like Naspers for OLX or Uber’s investors), keep investing in growth</p><p>[1:42] <br>with losses</p><p>[1:42] <br>Especially if the prize is big enough at the end</p><p>Jay Farber [1:43 PM] <br>i.e. Amazon not generating profit for 15 years</p><p>Fabrice Grinda [1:43 PM] <br>Yup</p><p>Neeraj Singhal [1:43 PM] <br>uploaded and commented on an image: BpfMqKzCUAAL7Xw.jpg <br>1 Comment<br>Now we’re talking! I want to point to this and offer a thought</p><p>Neeraj Singhal [1:46 PM] <br>Subsidies and discounts make this cycle spin faster… but there is a ton of ‘inefficient spend’ that gets lost along the way. The more you subsidize and discount, the more your customers get used to depending on those discounts to ride. Today’s customer for ridesharing has no loyalty to one brand, so I’d argue there may always be a market for another player. People have Uber, Lyft, Via and Juno on their phones in New York… and who over is cheapest will get the ride. Where this all falls apart is when the rug is pulled from under you, and you have to have an efficient enough operating model and decent enough algo to continue your business without the discounts.</p><p>[1:50] <br>To <a href="http://twitter.com/regularnick">@regularnick</a> ‘s point earlier (and he was formerly at Lyft for context), EBITDA is overlooked in the short term to try to tackle market ownership in the long term, but I’d argue some players won’t last in the absence of those subsidies. Ola vs Uber, and FlipKart vs Amazon are prime examples of massive spending, with an eye on the long-term… but not sure the players will last is they haven’t built and designed their businesses around the absence of subsidies and investor gravy.</p><p>Nick Greenfield [1:55 PM] <br>to your earlier point <a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> ‘s, is there’s a difference between vertical ownership of the assets in the marketplace that ultimately give you long term EBIDTA growth? Amazon has invested heavily on fulfillment centers, delivery solutions, etc. At what point should a marketplace think about becoming more vertically integrated? Post liquidity? Day 1?</p><p>[1:57] <br>Airbnb for example has to be thinking about this now as their ability to add supply will at some point hit a wall, even with low rates of churn.</p><p>Allan Grant [1:59 PM] <br>Guys — thank you for the great conversation! Unfortunately I have to jump off, but if anyone wants to continue chatting on any topic, please private message me or <a href="http://twitter.com/mention">@mention</a> me in this room, and I’ll jump back in later tonight.</p><p>Josef Feldman [2:01 PM] <br>Thanks <a href="http://twitter.com/allan">@allan</a> <a href="http://twitter.com/fabricegrinda">@fabricegrinda</a> <a href="http://twitter.com/dkalt123">@dkalt123</a> <a href="http://twitter.com/ronen_lazar">@ronen_lazar</a> great convo!</p><p>[2:02] <br>And thanks <a href="http://twitter.com/neeraj">@neeraj</a> for moderating. We’ll be doing another marketplace related chat in a few weeks.</p><p>Fabrice Grinda [2:07 PM] <br><a href="http://twitter.com/regularnick">@regularnick</a> I am not sure I understand the question, but 2 ways I think of verticalization: <br>1. In classifieds, we try to win C2C for sale first because it’s the category where people recur (come twice / month on average) as opposed to real estate or cars where it’s once every 5 years. So if you win horizontal c2c you have a platform to then verticalize and win car and real estate c2c and then b2c. That said you first need dominance in c2c for sale before doing any of that.</p><p>2. If you mean “offer more services” and “do more work” and extend the value chain it really depends on the vertical. At the end of the day you want to be a marketplace and not take inventory, own the supply etc. Now OLX can add optional payments and work with logistics providers but we don’t want to actually do it ourselves for instance</p><p>Nick Greenfield [2:09 PM] <br>Thanks! That answered my question!</p><p>Fabrice Grinda [2:11 PM] <br>Bye guys! You can find me at [redacted]@fjlabs.com if need be</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=df1bfbc95467" width="1" height="1" alt=""><hr><p><a href="https://medium.com/venture-capital-think-tank/1-marketplaces-chat-w-founders-of-olx-fabrice-grinda-hired-allan-grant-inturn-ronen-df1bfbc95467">Marketplaces Chat w/ Founders of OLX (Fabrice Grinda), Hired (Allan Grant), Inturn (Ronen Lazar)</a> was originally published in <a href="https://medium.com/venture-capital-think-tank">ThinkTank.vc</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>