Thinking Machines Builds AI for Human Collaboration

Thinking Machines has revealed its first big idea: AI that works less like a chatbot and more like a collaborator.

On Monday, the company founded by former OpenAI CTO Mira Murati said it is developing what it calls “interaction models.” The concept, according to the company’s description, centers on systems that let people collaborate with AI in more natural ways. Reports indicate those models continuously take in audio and video, widening the flow of information beyond typed prompts and text replies.

The company is betting that the next step in AI will feel less like issuing commands and more like working alongside a responsive partner.

That framing matters because it signals a push beyond the now-familiar chatbot interface that has defined much of the generative AI boom. If Thinking Machines follows through, its work could point toward systems that interpret speech, visual context, and ongoing human activity in real time. Sources suggest the ambition is not just smarter answers, but smoother cooperation between people and machines.

The announcement remains narrow on product details, timelines, and technical specifics. Still, the message itself offers a sharp clue about where the company wants to compete. Rather than chase attention with broad claims, Thinking Machines has put a label on a specific vision of AI: systems that stay aware of more of the world around a user and respond in a more continuous, human-centered way.

What comes next will matter far beyond one startup. If interaction models become viable, they could reshape how people use AI at work, on devices, and in everyday decision-making. For now, Thinking Machines has offered a blueprint, not a finished product — but in an AI race crowded with similar promises, a concrete direction already stands out.