Better Training Data Better AI.
Today the limitations of this paradigm are clear: significant efforts for marginal performance improvements, restricted access to overspecialized engineers, low explainability, low control, and prohibitive project costs.
Since the 80's the AI paradigm has been Better Models = Better AI.
Today the limitations of this paradigm are clear: significant efforts for marginal performance improvements, restricted access to overspecialized engineers, low explainability, low control, and prohibitive project costs.
@Kili Technology, we are believers.
3 years ago, Edouard d’Archimbaud and I started Kili Technology to change the terms of the equation.
We devised a method to radically improve performance, return control and efficiency to businesses over their AI and unlock companies’ digital transformations.
At that time, very few understood our mission. The hype within machine learning teams was either to fine-tune hyperparameters or to test the latest models’ architecture, with ML teams spending 80% of their time on data preparation. Very few projects made it to production. Data labeling was considered the least value-adding activity in the AI value chain.
AI is not deterministic: if you repeat the same experience, you can have 2 very different results. Optimizing the process by changing the parameters of the model or the model itself only adds an extra layer of fluctuation.
If your car no longer starts, you either replace the engine or fill it up with the right fuel instead.
Selecting the right data, ensuring that it is unbiased, and iterating more quickly on the model's output to bring in additional data until the expected performance is achieved can increase success tenfold.
Better Training Data = Better AI.
@Kili Technology, we are doers.
Edouard and I left our comfortable jobs to start (again :-)) this new venture. We wanted to build a robust company (strong value proposition + solid business model), so we didn’t raise any money until we had very clear signs of strong market fit. We self-funded the 1st year of our venture by doing some consulting work. We were then able to benefit from a lot of debt to focus on our product (France is an amazing country to start a business Thank you #BPI). 1st lockdown has been tough as all ongoing commercial deals suddenly froze. Slowly we manage to build an amazing team @Pierre 1st employee, @Maxime @Phillippe @Paul. A team of doers, following a clear vision, developing an excellent product, dedicated to one success: our customers' success.
End of 2020 we closed a $7M seed round with Serena @Marie @Bertrand & Headline @Jon to have the resources to build an ambitious company. Since then, the size of the team has doubled, always setting the bar higher.
Today, just six months after this seed round, we are announcing a $25M Series A with Balderton Capital @Bernard to build a global leader. We were obviously not in need. This is the opportunity to go faster.
Every customer is a new step.
So is every new person joining the company?
The new AI’s paradigm is Better Data = Better AI.
Kili Technology helps enterprises create and manage training data, in order to accelerate AI projects.
We work with hundreds of customers in France, Europe, the United States, and Asia.
We are hiring 70 people in the 12 coming months.
If you share our ambition, join!