Chai Research moves chatbot company from Cambridge to Palo Alto

Chai Research, a start-up which has developed an interactive chatbot platform which already has 100,000 daily users, has relocated to Palo Alto in California, saying that the weak pound and a low-investment UK environment has played a big role in its departure.

Chai Research was founded by a team of five colleagues who all met at the University of Cambridge. Their first venture was Seamless Capital, a financial trading platform. Gradually, however, their skills – the team includes mathematicians and an astrophysicist – evolved and the quintet set up Chai Research.

The resulting ‘Chai’ mobile app allows users to play with a powerful new class of machine learning models “capable of writing, drawing and creating with credible, sometimes superhuman results”.

The team raised funding from a prominent American billionaire at the start of the year and co-founder Thomas Rialan, who left Cambridge for California earlier this week, says the investment mood music is immediately apparent.

“I already had a former CEO of GitHub offer $250k on the basis of a 15-minute phone call,” he said, speaking from Palo Alto. “The due diligence here is much less, and the optimism and appetite for risk is much higher. If you want to start a technology business or launch an app then it has to be California.”

Thomas’ Californian colleagues are William Beauchamp, CEO and co-founder, Rob Irvine, CTO, Joe Nelson, COO, and Tom Xiaoding Lu, machine learning researcher.

Thomas says of the origins of Chai Research: “While we were working together at Seamless Capital, we said among ourselves ‘this is going very well, so maybe now can we build something 10 times as exciting’.”

He clearly believes in the platform’s potential, while acknowledging that the team is “figuring out the tools needed to build a really good bot”, adding that the chatbox format whose time is on the cusp of huge success, he says.

“You might want to book a holiday weekend on the Cotswolds, or you’re stuck on your homework... it’s any type of interaction where the medium of information is language, so that’s what we’re doing – building tools to make compelling chatbot experiences.

“The core thing is that the chatboxes – there’s millions of them – get to know you. You can speak to anyone but the chatboxes learn what you like, they get to know you in the same way as Google or YouTube. So in future, instead of going to Google or YouTube, you’ll go to one of these chatboxes.”

The revenue stream comes half from subscribers – you get 70 free interactions with the bot before you need to go premium – and half from advertising.

“We’ve had millions of engagements,” Thomas adds. “There’s around 100,000 users every day, with 300m messages being sent to chatbox every month. It’s mostly teenagers – half are in America, the other half is spread evenly across the world.

“We currently are making over $1m a year. The plan for the next six months is that we’re hiring as many people as we can afford to, to make the chatboxes more useful in terms of information – they are mostly being used for entertainment at the moment – and smaller in terms of memory usage.”