Hack Trend

Investor Pitch Workshop at GA - thanks for coming!

Thanks to everyone who came to the Investor Pitch Workshop event at General Assembly in NYC last Friday 3/30.  It was a full house, and a fun happy hour afterwards!  Here are the companies and founders that presented:

• Metropolist - JC Goodrich
• Peer Belt - Krassimir Fotev
• Screach - David Weinfeld
• Paperlex - Alison Anthoine
• TigerTrade - Tanjila Islam
• BusyFlow - Jaro Šatkevič

And thank you judges for your valuable feedback:

• Weston Gaddy, Bain Capital Ventures
• Alex Goldberg, Canary Ventures
• Kamran Ansari, Greycroft Partners

DARPA’s call for humanoid robots

Interesting announcement by DARPA that they will fund six hardware teams and twelve software teams in the Grand Challenge to design a humanoid robot. This is the extension of the same challenge for the autonomous driverless cars that we’ve all heard of.  

The requirements for the humanoid robot are:

• Navigate itself into a open-frame utility vehicle, hop into the driver’s seat, and drive the vehicle to a specified location.
• Exit the vehicle, unlock a door, and go through the door.
• Safely travel down a 100 meter-long hallyway littered with debris.
• Climb a ladder
• Fix a gas-leaking pipe
• Replace a broken pump

I wonder if there are requirements on battery power, speed, toughness, etc.  More on this from VentureBeat.

Slides from GA talk

Here are slides from my General Assembly talk on product development for entrepreneurs on 3/28/12:

Factors for Success in Consumer Internet

Over the past few weeks, I’ve enlisted the help of a few MIT Sloan Master of Finance students to take a look at the consumer internet space and identify high level factors for success. In the next post, I will share some of those findings.  

But to start, here are a few areas that I think are important drivers.  Let’s see how these stack up against their analysis for the next post.

• Is viral built into the product i.e. users are incentivized to share with other users, resulting in non-linear user acquisition. Are there network effects, i.e. single sided, cross platform, both etc – each user that join makes platform more valuable to other users or other side of platform i.e. advertisers, and more users makes it more difficult to leave (higher retention). Social networks are inherently viral.  Daily deal sites like Groupon have designed viral into the product (share with X number of people to activate the offer or receive a further discount).
• Related to above point, certain models built around helping consumers store data like Shutterfly and one of our companies SpringPad, the more you store the more you have to stay on the platform because leaving means you abandon all your data.  Sharing then multiplies this effect as more people become participants and are dependent on this data.
• User engagement – how often user comes back to the site – which then drives monetization, either ad-based (media model), ecommerce, virtual goods, or subscription. Google isn’t viral but it’s a better mouse trap for search, and once you use it and get good results, you use it all the time, and every search query is a proxy for user intent so it’s highly monetizable via targeted advertising.  Another example would be addictive games like Angry Birds monetized through ad views and the mighty eagle in-game upgrade.
• Distribution/buzz, getting access to the right channels, partnerships. This includes factors like celebrity power i.e. Ashton Kutcher putting Twitter on the map.  Influential angel investors also serve as signals to the market. Examples in the Bay Area include Reid Hoffman, Peter Thiel, Ron Conway, Kevin Rose, Ashton Kutcher (who’s also an investor) etc
• Input / output value ratio.  Low friction to participate and get started and low effort for users to get value out of the product. The value can be of various forms - it can be entertainment, discovery, discounts, news, productivity enhancements etc.
• The factors above are interdependent and drive each other. Generating PR buzz with viral loops built into the product extends the effect of the PR. Each new user brings additional users through sharing content, driving higher engagement and monetization opportunities.  If the content shared is valuable and need to be stored/accessible everywhere, this again creates stickiness and increases engagement, which in turn drives more users through sharing etc. High value to input ratio ensures that users actually complete the steps in the product and get to the “share” button at the end, in addition to encouraging repeat usage.
• High value user demographics i.e. female audience, and product categories such as luxury/fashion – recent examples include Gilt, Rue La La, Beyond the rack, Shoedazzle, One King’s Lane etc
• New channel/device.  We’ve seen the movie repeat itself.  New businesses are built (many times adapting past successful models) each time a new channel, device, mode of interaction emerges. Evolution of channels/media: newspaper, radio, TV, web, social media etc. Interaction: text, rich media (video), voice, touch etc.  Device: TV, game consoles, desktops/laptops, mobile, sensor-enabled devices.  This then creates b2b2c opportunities for sectors like advertising, as advertisers need new ways to reach consumers who are now spending time and engaging on different channels and devices.
• The anti-X.  In the social media world of over-sharing and hosted centralized world where individual ownership and controls is reduced, could there be opportunities to attract the fraction of population that don’t want to participate?  Does this create opportunities for security, privacy solutions?

How web/mobile is changing consumer healthcare

After returning from the Partners’ Connected Health Symposium this past week, I have a few thoughts on how web/mobile/social media is changing the way consumers interact with healthcare:

• Social media - patientslikeme already demonstrated that patient-reported data from a user base that is highly engaged is valuable, and that there can be a real business model from data analytics, clinical trials recruitment etc.  What made it work?  First is engagement. Chronic disease patients are more incentivized to learn about and improve their conditions, because they identify with it and therefore are more active in these online support communities.  Second is high patient value, and therefore higher value lead-gen opportunities with pharma and researchers who want to target those patients.  Third is proprietary data and analytics that can be created, such as ability to analyze whether certain medications actually made an impact on treating the disease, adjusting for the natural progression of the disease.
• Mobile / sensors - GreenGoose takes an interesting approach of making wireless sensors very cheap and in the form of stickers, so they can be applied to physical items such as toothbrushes, bicycles, frisbees etc to allow passive tracking of activity data.  Passive is the key, it doesn’t all have to be coming from smart-phone based apps, it just has be easy and low friction for users.  Fitbit is another example, offering a device+subscription model to help users track, analyze and compare their activity levels.  Zeo is doing the same, but for sleep - they call themselves the WeightWatchers of sleep.  Mobile and sensors adds granularity to data which creates more frequent engagement that in turn drives greater value in consumer applications.  There was a talk during the conference by Rosalind Picard who founded Affectiva, an MIT Media lab spinout that uses a device called the Q sensor to measure skin conductance which is an emotional indicator — it turns out that the data collected by the device can be used as a real-time predictor of the onset of seizures.  Separately, Picard’s research group has also developed an algorithm for measuring vital signs of a person such as heartbeat through just an ordinary web cam, bypassing body sensors entirely.
• Making information available to consumers - very simple concept, but still not the norm in healthcare.  Starting with doctor’s appointment times online, which ZocDoc has gotten traction in offering to consumers - the business model there is bringing new patients to doctors and helping doctors optimize scheduling of patients (and thereby optimizing revenue / reducing lost revenue).  How about cost information around health procedures and doctor visits?  Castlight is doing it from the employer-sponsored health plan angle.  160M people in the US are covered via their employers’ health plans, so employers are an important stakeholder as well as a go-to-market channel.  As the world is moving more towards high deductible plans, consumers will be more incentivized to shop around and price compare when it comes to healthcare.  lifeIMAGE is applying cloud data storage (dropbox-style) to medical imaging, allowing patients and physicians to access X-Ray, MRI etc images online - the interface is simple, it looks and feels like email.  Integrating hospital ratings, performance data (i.e. mortality rates), doctor ratings, etc will become standard feature in many consumer facing health apps.

TechStars NYC demo day October 2011 - Trends and Patterns

I went to TechStars demo day in NYC yesterday (Tue 10/18).  From what I saw, I’m increasingly convinced that we are now in the second wave of the new consumer web because we are seeing verticalization, aggregation, and normalization of models that were successful in the first wave.  Examples:

• Verticalization - Contently is a marketplace for brands to hire professional writers.  This is a verticalization of elance / odesk / craigslist.  Similarly, Coursekit which is creating a social network for education, is a verticalization of Facebook.  Ordr.in is a verticalization of ad networks (yahoo, google) and lead-gen models applied to restaurants.  As first wave consumer web services get big, they accustom the consumer to a type of usage behavior and interface which creates market readiness for these types of verticalized models.  The other thread here is the application of successful consumer web models to disrupting industries which have been resistant to change due to the particular industry structure.  In that case, what determines the success of these types of new services will be how they go to market and get adoption, such as circumventing established buying hierarchies (education is a good example).
• Aggregation / normalization - As more and more similar consumer web offerings come online, opportunity is created for aggregation of these services.  One function or value of aggregation is normalization which is a way to perform an action across multiple services, abstracting away the particularities of each service.  Examples of this include Dispatch which integrates across file sharing services that give users a drag-and-drop way to collaborate across these services.  This model can be extended to other actions like scheduling and time, where Spontaneously is focused so that regardless of your calendaring system there should be one place to figure out when to do what with your friends/colleagues.
• The next iteration of X - the next generation of an existing technology or usage model by incorporation of new data types, devices, etc leveraging major trends such as mobile, rich media etc.  One example is Piictu which is basically group chat with images (the next iteration of chat) - users post interesting photos around certain topics and other users can respond with photos.

The demo day show also included a speech by Mayor Bloomberg who was predictably and justifiably optimistic about New York’s startup scene.  An interesting stat he mentioned is that NYC has more undergraduate and graduate students than Boston has people.  Being a Bostonian and a former New Yorker, I was still a bit shocked with that stat so I looked it up. Well it was close: 594K university students in NYC vs 620K people in Boston city proper.  In the greater Boston area there’s about 330K university students (out of a population of about 4.5M).  This makes the ratio of university students to total population for both cities at about 7.3-7.4%.  At the end of the day, NYC just has a lot of people (8 million) at high density which makes it a great test market for consumer internet startups.

Managing software product development (for entrepreneurs)

Thanks to all the entrepreneurs who came to my talk on managing software development at the Harvard iLab on Tue 10/11/11.  Here are the slides from my talk:

[3/28/12] Note: updated version here

Software Project Management for Entrepreneurs

Why it’s great to be in Kendall Square

Kendall Square in Cambridge has become quite the hotbed of innovation.  It’s a great place to be if you’re a startup, or a stakeholder in the entrepreneurial ecosystem.  Here is why:

Academic Resources

Being close to MIT has many benefits, including access to the incredible number of research labs, including but not limited to:

• Media Lab - one of the most unique labs in the world where digital media, computer science, and physical world (i.e. making stuff) intersect.
• Stata Center - home to MIT’s many computer science and artificial intelligence research groups (link to research groups)
• Koch Institute for Cancer Research - a brand new research lab focused on cancer research, opened in 2010.
• Brain and Cognitive Sciences Complex - shedding light on the least understood area of the human body, the brain.
• Broad Institute - molecular and genetics research, including study of inherited and infectious diseases and cancer.
• Many more corporate research labs as well, all making Kendall Square the most interesting research mecca of the world (certainly for life sciences)

Startup Resources - incubators, office space, events, etc

• Cambridge Innovation Center (CIC) - flexible, modern office space for startups.  Venture Cafe hosts regular open networking events for the local entrepreneurial community.
• Dogpatch labs - Polaris Ventures’ incubator providing free office space to startups. 
• TechStars Boston - an incubator/accelerator program for startups, offering office space, mentorship, and seed funding.
• Microsoft NERD (New England R&D) - Microsoft has positioned itself as a startup-friendly hub, hosting many regular entrepreneurial events.
• MIT Enterprise Forum of Cambridge - hosts many technology and startup events, some on MIT campus.

Venture Capital Firms

• Fairhaven Capital - where I work and one of the first to be in Kendall Square!
• Bessemer Venture Partners
• New Atlantic
• Atlas Ventures
• Flagship Ventures
• Google Ventures
• CommonAngels

Large Companies

• Google and Microsoft are both in Kendall Square.  Microsoft has a nice space on One Memorial Drive overlooking the Charles River where a lot of startup events are hosted these days.
• Other IT companies include IBM, EMC, Akamai, Hubspot

Hangouts - coffee, beer, food

There are now growing number of locations to hang out, grab a coffee/beer and/or a bite:

• Voltage Coffee - where many entrepreneurs and VCs have their meetings now.
• Abigail’s - one of the newest restaurant/bar in Kendall Square, oyster bar, good cocktails.
• Meadhall - 140+ beers on tap and great food, 2 stories.
• Cambridge Brewing Company (CBC) - a tradition, all beers served here are brewed by CBC, food is pretty decent too.
• Flat Top Johnny’s - affordable beers and pool, one of the few choices in the area for hosting a corporate event.
• Friendly Toast - great for breakfast anytime and a coffee, outdoor seating is nice when weather is reasonable
• Hungry Mother - gourmet take on southern cuisine, the winner of the 2011 Munch Madness. 
• Lord Hobo - great selection of beers and food (especially the chicken drumsticks).
• Mulan - one of my favorite chinese restaurants in Boston. Taiwanese style food, quite authentic.
• The food trucks at MIT, especially Clover

Integration of data to create new value

When I worked at Microsoft in the early 2000, the company was still forming the vision for .NET and that one day the online world will be connected by XML data pipes with apps communicating over SOAP and web services.  Today, the web continues to be more and more fragmented with some web applications offering great APIs and/or RSS feeds – some web sites still do not and sometimes it’s not in their best interest to make it easy to export their valuable data.

There is a growing trend of startups asking users to give their log-in credentials to multiple data sources so data can be pulled into one place via an API or via screen-scraping.  The results are great services like Mint.com which analyzes the users’ disparate financial accounts (bank, credit cards, etc) for potential savings to arrive at insights which would’ve been difficult without an aggregate view of the user’s data.  What other sources of user data would be valuable to aggregate?  I can think of a few:

• Social network – Facebook, Twitter, LinkedIn, Foursquare etc. This can be done mostly via APIs.  Rolling up the user identity across these networks into a master ID is already supremely valuable.  Event data, any type of content like trending news, location data, social graph, influence scores etc can all benefit from a broader view of a users social media data.  There are multiple business models here, both servicing the user (social discovery, information, entertainment, dating, etc) or third parties like brands willing to pay for this information as a channel to reach users or for market data.
• Travel – booking information from all different sources including airlines, trains, buses, hotels, rental cars etc. TripIt does a great job of aggregating this information into a coherent itinerary organized by date, location and sells premium features like real-time travel alerts. What’s even more interesting is that since all the travel information is sent to users via email anyways, TripIt only needs to integrate with a user’s email (asking for only 1 login credential) to pull in all the relevant information.  I wish there’s a better way to manage my frequent flyer programs and other travel loyalty programs — what if there’s an agent that aggregates all the info about my programs and then helps me book the right type of trips with the right airlines/hotels/trains to maximize my points?
• Daily deals –consumers are getting deal fatigue from an extraordinary number of daily deal sites but since deals are all public, aggregators such as Yipit can pull in that information and then target the right offers to the right users, without the need to ask for their login credentials to any of these sites.  I just wish that I can organize the deals I’ve already bought easier and all in one place, since I’ve already had a few expire on me.  Or what if I can snap a photo of my receipts or forward my electronic purchase histories to one location, and get coupons and recommendations.
• Healthcare – hospital bills, insurance claims, employer health benefits, health savings accounts, etc. bringing this data together to give consumers a view of their healthcare spending and potential for savings, mashed together with provider information like hospital and doctor ratings.  Companies like Cake Health and WellnessFX are taking a stab at this, and ZocDoc already has info on doctor availability and what insurance they accept.
• All my bills – credit cards, cell phone, cable, internet provider, rent, utilities, mortgage, education, etc.  One place to go to pay my bills with a clear calendar of when the next payment is due, can help me save, plan and budget properly.  Lots of business model options here including doing a Mint.com-style lead gen.  BillShrink has done a good job on mobile carriers and is now moving to do cable, gas, and savings & CDs.
• Safety and emergency response – crime rates, neighborhood police reports, fire reports, weather, natural disaster information, local traffic and outages, etc.  This data can be mashed up with my real time location information to help me plan my commute and better respond to emergencies etc. Life360 has done a nice job with a mobile offering.
• Event and appointments – all my work, social calendars across Outlook, Google, Lotus Notes (?), Facebook, etc mashed up with events and location information.

Robots – When and how?

The idea of a general purpose intelligent robot is something we’re all too familiar with.  From the Jetson’s to the Terminator, science fiction has fascinated us with robots smart enough to accomplish tasks ranging from household chores to social engineering for the purpose of seek and destroy.  But when and how will robotics finally happen?

AI and machine learning specifically has hit a plateau in the last 20 years, accelerated only in recent years by the availability of massive amounts of data for which statistical methods are more applicable and more powerful – examples are plenty including Google and the recent jeopardy champion Watson.  Industrial robots represent a more near term reality because manufacturing environments are fairly structured.  The moment a robot is required to handle unstructured environments like the average home, the problem becomes extremely hard – the iRobot roomba is one of the few widely adopted consumer robots; that said, elements of intelligence have been making its way into our cars for quite some time now (collision detection, auto-parallel park etc).

But the question still remains: When will we have a general purpose robot smart enough to do our laundry and clean our house, and maybe bake an edible batch of chocolate chip cookies?  Are these scenarios always just 10-20 years away?  Maybe the future is finally on the horizon.  Check out these recent research breakthroughs using the PR2 R&D robot from Willow Garage:

• PR2 fetching beer and playing pool, and then cleaning upafter a party
• PR2 getting a takeout sandwich
• Berkeley taught the PR2 to fold laundry
• MIT taught the PR2 to bake cookies from scratch

Robotics today look a lot like what the PC industry did in the late 70s/early 80s.  Hobbists are using whatever off the shelf tools available to create ever cheaper and more intelligent robots.  There was even a contest a few years ago for building a UAV under $200 that can fly around a building and take off and land autonomously, built using cheap sensors and motors.  High school students across the US are competing in the US FIRST robot competitions; many of these students are now aspiring to “major” in robotics in college.  Now the PR2 above costs $400K, the largest cost on its BOM are the robotic arms, each costing over $100K.  Some innovative mechanical engineering needs to occur with the help of Moor’s law-like dynamics to bring overall costs down.  But hey we’ve seen that before in the PC industry right?

The latest wave of affordable robots are telepresence robots (Texai, Jazz, Luna, etc).  I think the value of those robots are limited because they don’t have arms, and therefore they are nothing more than a mobile skype/facetime terminal.  For a robot to be useful around the house, we need the following to happen:

• The entire robot needs to reach the price point of $5K or lower.  Robot arms (actuators) prices need to decline significantly.  That said, consider if a robot can save you 4 hours of housework a week at $25/hr, so in a year the value of that robot can be as high as 4 x $25 x 50 = $5,000.
• Motor skills need to improve i.e. ability for robot to go up and down stairs – or it’s a combination of that and future homes being more “robot ready”.
• Machine vision needs to get better to recognize household objects it has never seen before with reasonable accuracy — a cloud-hosted commonsense knowledge base would be helpful so robots can leverage crowd-sourced learnings of other robots.
• Development platforms are starting to pop up and gain momentum. The ROS (robot OS) and the corresponding app ecosystems, for example, needs an owner, a company to step up and take ownership of a core set of apps with quality guarantees and a nice consumer-ready “user interface”. This would be analogous to Microsoft DOS (ok not exactly), IBM Linux and maybe more fittingly Google Android.

So, there are some very basic computer science challenges like machine vision and “common sense” that need to be overcome which makes estimating the timeframe of robots difficult.  It’s also hard to directly apply the PC history analogy i.e. the invention of the GUI and mouse was a paradigm shift but machine vision is a totally different problem.

But we are closer than ever, and we just need enough engineers working on robots to overcome the last few challenges and push robots into the mainstream home.  I wouldn’t bet against a general purpose robot happening in 20 years.  Terminator insurance, anyone?