Microsoft has acquired conversational AI startup Semantic Machines for an undisclosed amount. Founded in 2014, the startup’s goal was to build AI that can converse with humans through speech or text, with the ability to be trained to converse on any language or subject.
Semantic Machines has raised about $20.9 million in funding from investors, including General Catalyst and Bain Capital Ventures.
In a blog post, Microsoft AI & Research chief technology officer David Ku wrote that “with the acquisition of Semantic Machines, we will establish a conversational AI center of excellence in Berkeley to push forward the boundaries of what is possible in language interfaces.”
Microsoft is also acquiring some key talent in the deal: Semantic Machines’ CTO Larry Gillick, who was previously Apple’s chief speech scientist for Siri; as well as AI professors Dan Klein and Percy Liang, from UC Berkeley and Stanford, respectively.
While Microsoft’s conversational chatbot Tay was a disaster in the United States, an extremely similar chatbot called Xiaolce has been a resounding success outside of the US, racking up 30 billion conversations with more than 200 million users worldwide. Microsoft’s virtual personal assistant, Cortana, also lives in every Windows 10 computer, though it’s not as useful in a voice-only setting like a smart speaker.
Semantic Machines is believed to fit in well with Microsoft’s conversational AI-based products. These include Microsoft Cognitive Services and Azure Bot Service, which the company says are used by one million and 300,000 developers, respectively, and its virtual assistants Cortana and Xiaolce.