Thinking Machines, founded by former OpenAI CTO Mira Murati, announced development of 'interaction models' designed to enable real-time AI collaboration through continuous processing of audio, video, and text inputs. This represents a shift from current AI models that operate in single-threaded mode, waiting for users to complete input before responding.
Mira Murati's departure from OpenAI to launch Thinking Machines signals continued talent fragmentation in the AI industry, with senior technical leaders pursuing independent ventures. The announcement of interaction models addresses a fundamental limitation in current large language models: their inability to perceive and respond to real-time multimodal input streams. Today's models operate reactively, processing complete user inputs in isolation rather than maintaining continuous awareness of user behavior and context. This architectural shift toward simultaneous audio, video, and text processing could fundamentally change human-AI interaction patterns, moving beyond turn-based dialogue toward genuinely collaborative workflows. The technical challenge is substantial—maintaining coherent reasoning across continuous input streams while managing computational efficiency requires novel approaches to attention mechanisms and state management that current transformer architectures weren't optimized for. For the broader AI ecosystem, successful interaction models could accelerate deployment in applications requiring real-time responsiveness: customer service, collaborative software, and autonomous agents. However, the competitive landscape is crowded; major labs including OpenAI, Google, and Anthropic are exploring similar multimodal capabilities. Thinking Machines' success depends on achieving meaningful differentiation through superior reasoning quality or efficiency. For developers and enterprise customers, interaction models could unlock new use cases previously impossible with existing chatbot interfaces, potentially creating new market categories. The announcement itself carries limited immediate market impact but signals accelerating AI capability development and the continued exodus of top talent to create specialized AI startups focused on specific technical problems.
- →Thinking Machines is developing 'interaction models' that process audio, video, and text continuously rather than waiting for complete user input
- →Current AI models operate in single-threaded mode with no perception of user behavior until input is finished
- →The innovation addresses real-time collaboration needs that existing large language models cannot satisfy
- →Mira Murati's startup joins a crowded field of companies pursuing multimodal AI capabilities
- →Successful deployment could enable new applications in customer service, collaborative software, and autonomous agents