- The computer vision startup Tractable “sees” and analyzes the costs of repairing cars damaged in accidents.
- The British startup is now worth $ 1 billion after raising $ 60 million.
- More computer vision unicorns are expected, but experts say barriers to mainstream adoption persist.
Tractable says it is the UK’s first computer vision unicorn, having raised $ 60 million in new funding at a valuation of over $ 1 billion.
The round was led by existing funders Insight Partners and Georgian.
Company chairman Adrien Cohen said the company increased revenue to eight figures, thanks to large insurers starting up as clients last year.
The London-based startup, founded in 2014, has primarily applied computer vision capabilities to assess damage to cars after an accident. It partners with insurers to help make an initial assessment and estimate repair costs, which can reduce the time a car spends in the body shop. The company has trained its algorithms on dozens of photos of damaged cars and claims the system is as accurate as a human.
Tractable was co-founded by computer scientist and former hedge fund quant Alex Dalyac, machine learning specialist Razvan Ranca and former Lazada executive Adrien Cohen.
“Reaching that milestone isn’t important in and of itself, but that’s what it says about the impact and scale of our technology, the validation of reaching that scale,” Cohen said of unicorn status. of the company. The startup has around 20 insurance clients in the United States, Europe and Asia, including Geico, a subsidiary of Berkshire Hathaway.
Although initially specializing in auto repair appraisals and estimates, the company is now expanding into property damage analysis and even car purchasing.
“We’re going to dig deeper, we think our AI can handle the cases where you want to inspect the condition of a vehicle, not just in the event of an accident, so when you buy, sell or lease,” Cohen said. “All of these events, where you can speed up the process by understanding the condition of the vehicle from photos.”
In theory, the platform could partner with a used car platform like Cazoo to assess the condition of a car put up for sale. Cohen said car rental companies and automakers are also potential customers.
Asked about revenue growth, Cohen said the company was privately held and would not reveal specific numbers. “It’s an 8-figure turnover [number], with 600% growth in the past 24 months, ”he said, adding that the company had only raised $ 55 million in outside capital before the new round.
Computer vision startups and the road to commercialization
Tractable is part of a wave of startups benefiting from the maturation of computer vision. According to this year’s edition of the AI annual index, collated with Stanford University, computer vision is becoming more and more “industrialized”.
Alessio Petrozziello, a machine learning research engineer at London-based data mining start-up Evolution AI, says that, more generally, computer vision has a few hurdles to overcome before it becomes fully integrated.
“There is certainly a push to bring these models to market, but it is clear that they are not at the level where you can [fully] count on them, ”he said. “For example, for an autonomous car, it cannot make any mistakes, certainly neither can a human. a model, and the model makes a mistake, who is responsible? There is no clear answer. “
Eleonora Ferrero, COO at Evolution AI, added that the success of startups like Tractable is as much about execution as it is about core computer vision technology.
“Their bringing to market consisted of partnerships with key insurance companies that provided data, that’s an advantage,” she said, adding that Tractable had been smart to identify something that insurers were looking for. – increased operational efficiency.
Karen Burns, founder of the Fyma computer vision platform, said adoption depends on preparing customers for the technology. Fyma’s platform, formed on anonymized data, analyzes what is happening in a physical space – whether it is a business that tracks the movements of its autonomous robots for security; or a retailer measuring footfall.
“Before you can adopt AI, you have to go through a big transformation,” she said.
Tractable’s Cohen agreed, saying the company had relied on the quality of its AI development, but also sold the utility of AI to corporate clients. “A great playbook that we have discovered is how to deploy and capture the value of artificial intelligence in a business context,” he said. “It’s very difficult, and something we had to figure out.”