Coursera

Some interesting details on Coursera today:

- 6 million users, 108 partners (universities and training organizations), team size of 80, 550 courses

- revenue mostly from signature trek courses (ASP of $50)

- overall completion rate low (<5%), vs. higher completion rate for sig trek users (60%+)

- course instructors are suggested to budget ~400 hours to construct a course

- largest MOOC now (in terms of #users) and largest collection of education data

- many A-B tests done, on email communication, video format, grading, etc.

- feedback mechanisms include: multiple choice, peer grading, machine grading (numeric probs)

- signature trek homework verification: web cam and typing pattern. The participating institutes share revenue

- international expansion: Chinese now (through NetEase, a local partner that helps host content in China), and French / Spanish to come

Coursera team also has deep social root - the motto is to do what's good for students and it rolled out a financial aid program when starting the sig trek program. When asked about why Coursera is a for-profit company, the founder said that it was mostly due to funding. It managed only to raise $150k when it was trying to be a non-profit, vs. tens of millions as a for-profit.

What's really interesting about Coursera is its data on people's learning habits. So far, Coursera is only utilizing a small tip of the data, and conducting analyses on the surface. There is certainly huge potential in the future. 

Cortica - Image Recognition

Image processing technology has become a focus of M&A lately. Pinterest just announced its acquisition of a 2-men startup, VisualGraph, which develops visual recognition tech. By using the tech, Pinterest hopes to better tag out and identify objects in pictures, to enhance its ability to advertise relevant products to the users. For example, if Pinterest knows the girl in a picture is wearing a skirt, it would provide ads on skirts, vs. coats. The two founders are both technical, with experience in machine vision. One is a ex-googler, who built the early version of vision recognition system for Google in 2004. The other founder is a current Stanford MS student, who interned in Palantir, Facebook and Google. The advantage of this startup is its algorithm. It claims that it achieves "8x speed" of object recognition systems and is on par with Facebook's face detection capability.

Another startup that was acquired was LookFlow. Yahoo bought the 5-people team in Oct 2013, for flickr.  Yahoo also acquired another image recognition firm, IQ Engines, which focuses on text, object and people detection. LookFlow was founded in 2009 in Mountain View, on deep learning, to help with search.

Cortica, right now, is still an independent company. It was started in 2007, and has raised $7 million from Li Ka-Shing's Horizon Ventures and Venture Capitalist, Ynon Kreiz, as well as a recent $1.5 million from Russian tech group, Mail.ru. The company's key strength lies in its Image2Text technology which can automatically extract the core concepts in images and videos, and map the concepts to keywords and texts. You can find a demo of its tech on its website.  The company is promoting "in-image" advertising, which could embed / provide relevant ads associated with objects in images. This could be a very interesting area, given the ads and content organization baed on texts are already mature and we are just starting to exploit the vast number of visual materials available on our phones, internet, TV, and maybe Google Glasses. Turning images into texts is definitely one critical part of the puzzle to make things happen. The key questions include: 1) how good is the recognition and how much limitation there is (assuming it has to be a server-end web app?). 2) whether other tech will catch up and perform better before the company turns their IP advantage into something more sustainable - ads network  or else. 3) scale and growth of its monetization.

Cortica has offices in New York, Sunnyvale CA and Isreal. Contacts can be found on their website, and emails are below:

info@cortica.com
jobs@cortica.com

California, U.S.
440 North Wolfe Rd.
Suite WL151
Sunnyvale, CA 94085
Tel: 1.866.972.0972

Welcome to the Farm!

Welcome to the farm, and ... welcome to OUR farm. First, disclaimer: everyone is biased. So are we. We are three MBAs passionate about entrepreneurship. This blog is dedicated to share our observations and reflections on startups and entrepreneurship through our own lens colored by our own experience. 

What is the purpose of this blog? 

Three things have tremendously changed the landscape of entrepreneurship.

• Technology - things can be started in your garage now, thanks to computers, APIs, 3D printers, hardware sets, etc. 
• Infrastructure as broadly defined - internet coverage, smartphone penetration, even roads and cars for last-mile delivery. 
• Theories and mentalities - "fail early and fail fast", rapid prototyping, design thinking and lean startups.  

We are now in a world where thousands rise and thousands fall within a year, a month and a day. The incremental innovation cycles are becoming shorter, and the target markets of the startups are becoming more and more niche and fragmented. While we could follow the tech industry by reading about a few growing companies, like Google, Amazon and Ebay, in the headlines a few years ago, we are losing track, and overwhelmed by startup tsunami. For investors, entrepreneur-to-bes, MBAs, and early adopters, there is a need for frequent and thorough survey. Thus, this blog - to observe, learn and spot trends.

The blog also serves another purpose - data gathering to test a hypotheses. Some of us see startups scattered around a 3-D space along three axises:

• business model innovation - Birchbox, Zipcar and iTunes are examples of business model innovation
• technology innovation - Google search algorithm, iPhone and 3D printer rely mostly on technological / engineering inventions
• process innovation - Salesforce, Intuit and Evernote are more of products that enhance certain processes

Some hold the belief that as long as you are a great founder, like Elon Musk, you can do anything in any field. I have gradually started to realize that there may be certain startup-founder fit, which may determine the fate of the endeavor.  People are limited by their knowledge, experience and social network. Most startups started by MBAs in school are high on business model innovation, low on tech. And obviously, most engineers, esp. PhDs, start companies that apply their research findings, relying on tech innovation. Founders with industry experience, on the other hand, tend to know the pain points of their industry well, and start firms to improve processes. What type of team does it take to start a certain type of companies at a specific time? There may not be an answer. And if that is the case, it would satisfy our idealistic thinking that everything is possible. However, if there is a pattern, we want to know the reality and adapt. Thus, this blog - to understand startup - team fit better. 

So January 2014 would be the start. we will write about startups that we come across and read about. With tags, we will also try to organize them by sector and and type. One post for each company. Comments, referrals, coffee chat invites, .... all welcome.