Siri for all: turning speech into actionable data thanks to an alumni
Alexandre Lebrun (X94), Laurent Landowski and Willy Blandin just raised 3 million dollars. Based in Palo Alto, the three entrepreneurs have created Wit.ai, a start-up that offers developers a system to integrate a voice assistant in their application.
Founded in 2013, the start-up Wit.ai just raised 3 million dollars through the Andreessen Horowitz fund, one of the top venture capital funds in the Silicon Valley. Wit.ai's speech recognition technology allows all developers to add a voice control system to their application, be it for a mobile app or a connected object.
Interview with the founder and CEO of Wit.ai, alumni Alexandre Lebrun (X94)
How did you get the idea for your start-up?
Before Wit.ai, I had created a startup named VirtuOz that develops "Siri" software for the customer services of large companies such as eBay, PayPal, AT&T, SFR. These virtual agents worked well, but their implementation was very expensive because we needed to set up the sotware for each client.
After the acquisition of VirtuOz by Nuance Communications early 2013, I wanted to make "Siri" software accessible to everyone, including independent developers with no expertise in the field of natural language processing and artificial intelligence. This is the goal of Wit.ai.
Is there a need for such a software?
After cell phones, the next wave will be about connected objects, the Internet of Things: intelligent Pebble watches, Google Glass, connected car, smart house etc. We will soon be surrounded by objects without screen or keyboard. We will use and control those objects with our voice. To develop apps for such objects, the developers need a user-friendly platform that will allow them to set up their own "Siri" in a few minutes.
What is your business model?
Our product is based on automatic learning, the "Machine Learning". The process of learning is the same as for children: we must expose our platform to as many human interactions as possible. The developers who use Wit.ai thus also work to our benefit, because they teach the human language to our machines. In exchange, Wit.ai is free for them.
Large corporate companies that want to use Wit.ai on their own networks must pay.
How did you manage to raise 3 million dollars?
Before looking for investors, we launched an initial version of our product showing key aspects of our approach: the Minimal Viable Product (MVP).
This phase was critical because we had to move very fast. First because entrepreneurs start by using their own personal funds. Second because the market evolves very quickly: a product that was innovative 6 months ago might not interest anyone today.
We have opened the access to this MVP to developers by announcing its existence on Hacker News, without using any traditional marketing tool or spending a dollar. Success was immediate, as proven by the number of developers who were interested as well as the positive comments they made. At this stage, investors were already hooked.
Looking for investors too early on is one of the most common mistakes I see in young entrepreneurs. Often, the team does not really need funding right away, but they are looking for an external validation of their project. But ultimately, only the market can tell if a product is relevant or not. As much as possible, you should first try to confront your product to the market, even on a very small scale, then look for investors.
What are your plans for the future?
We are putting together a technical team in France, where engineers are much better trained on average than in the Silicon Valley. We would like to collaborate with the institutions I come from, École Polytechnique and Telecom ParisTech.
Can you tell us a bit about your team?
We are very complementary: each of us has a specific role. I define the product, Willy Blandin makes it and Laurent Landowski helps customers use it successfully. He manages as well all other aspects of our business (budget, human resources etc).
We are now a team of 11 people and we are still in this ideal stage where 100% of our resources are directed on the product and its users: no management, no politics, everyone moves very quickly towards the same direction. We can discuss matters very openly and take decisions very efficiently. As we grow, our next challenge will be to keep this spirit up as long as possible.
What qualities do you need to start your own business? What advice would you give to young entrepreneurs from Polytechnique building their own start-up today?
At the beginning of VirtuOz, as well as being CEO, I was also CFO, deliveryman, client advisor, telemarketer or plumber! It takes a lot of curiosity and an ability to embrace hundreds of tasks which may not always be glamorous with a smile. At X, we learn many things, but not necessarily how to cope to a hundred refusals when trying to sell your first product by phone.
To young entrepreneurs from X, I would strongly recommend to work with an experienced entrepreneur as your mentor, although I do not think that this mentor should be involved in strategic decisions (product, market...) because your vision should stay yours and yours only. However, this mentor can save you an infinite amount of time by helping you avoid the usual mistakes everyone makes at first regarding financing, hiring or managing business growth.