Mira Murati is building AI models for simultaneous real-time interaction

Mira Murati, former Chief Technology Officer at OpenAI, has founded Thinking Machines and introduced a new paradigm for AI interaction. The company is working on interaction models — models that perceive audio, video, and text simultaneously and respond in real time, mimicking natural collaboration between people.

New Interaction Paradigm

Interaction models are not simply new versions of language models. They represent a fundamentally different approach to how people interact with AI. Instead of writing a request and waiting for a response, a user can speak, share their screen, change context — and the model responds in parallel, like a conversation partner in natural dialogue.

According to Thinking Machines, current models perceive reality "in a single stream." While the user speaks or types, the model is passive — it sees nothing, hears nothing, understands nothing of what's happening. Processing only begins once all information has been fully received. This creates lag and unnaturalness in dialogue.

Interaction models must work differently: continuously perceiving the incoming stream, thinking in parallel with the user, and responding in real time. This is closer to how people actually converse — both participants listen, process information, and respond simultaneously.

What Changes

The shift from sequential to parallel interaction has significant implications:

• Naturalness — conversations with AI will become as smooth as talking with a person
• Context — the model sees tone of voice, body language, visual context, not just words
• Speed — there are no delays between a user's thought and the model's reaction
• Multitasking — the model can simultaneously listen, watch, analyze, and respond
• Adaptability — the system can notice when a user is confused and immediately help

History and Significance

Mira Murati left OpenAI in September 2024. Within a few months, she assembled a team and launched Thinking Machines with a clear goal: to redefine what "collaboration with AI" means. For her, this is not just another startup — it's an attempt to solve a fundamental problem with modern AI interfaces.

If Thinking Machines succeeds in realizing this idea, it could transform the way people work with AI across all domains — from professional work to education and creative pursuits.

What This Means

Interaction models represent the next step in the evolution of AI interaction. If current models are chatbots, then interaction models are nearly live dialogue. For users, this means more natural communication with AI. For business, it's the ability to create interfaces that don't require users to learn special syntax or commands.