Murati's first public appearance in 18 months signals Thinking Machines Lab is moving from stealth to market positioning, and her governance critique lands at a moment when AI power concentration is a live policy debate.
Key Takeaways
- First major media appearance in 18 months as Murati previews a new product category called “interaction models” at a Bloomberg event in San Francisco
- 200-millisecond processing intervals are the technical backbone of Thinking Machines’ approach to real-time audio, text, and video AI
- No release date given for the interaction models, framed as a first step rather than a finished product
- Murati said OpenAI would have “imploded” without her role during the five-day Sam Altman firing crisis in November 2023
- Researcher departures acknowledged but downplayed; governance and power concentration raised as a concern across the industry
What Happened
Mira Murati, CEO of Thinking Machines Lab and former CTO of OpenAI, made her first major public appearance in roughly 18 months at a Bloomberg event in San Francisco on Thursday, previewing a new product category the company calls “interaction models.”
Thinking Machines has spent the past year and a half largely out of view, raising capital, hiring researchers, and shipping one product: Tinker, an API for fine-tuning open-source AI models.
The interaction models Murati described are built to process continuous streams of audio, text, and video in 200-millisecond intervals. The goal is for the system to register the texture of real human communication, including interruptions, mid-sentence corrections, and pauses, rather than waiting for a finished prompt.
Murati declined to give a release date and framed the models as a first step toward a thesis the lab holds: that meaningful AI progress runs through closer human collaboration, not around it.
Why It Matters
Thinking Machines is competing for talent, capital, and mindshare in a market that has grown more crowded while it stayed quiet. OpenAI is a constant in the news cycle. Anthropic’s momentum dominates industry conversation.
And xAI has been folded into SpaceX ahead of a widely anticipated public offering, pulling in attention and investor interest of its own. Murati’s Bloomberg appearance reads as a calibrated move to remind the market that Thinking Machines is still in the race.
Skeptics have reason to wait and see. Murati offered no timeline, no pricing, and no customer count. She downplayed the recent departures of several high-profile researchers from Thinking Machines, attributing them to the compressed organizational volatility that comes with building a frontier lab from scratch.
That framing may satisfy some investors but it will not close the credibility gap that comes with being mostly invisible for a year and a half.
She also returned repeatedly to a governance concern that went beyond her own company. Too much weight, she argued, has been placed on the character of individual leaders and too little on structural checks. Good intentions, she suggested, do not substitute for accountability systems.
“When I wake up in the morning, I am not thinking about how to kill the competitor.” – Mira Murati, CEO, Thinking Machines Lab
Bottom Line
The next signal to watch is whether Thinking Machines puts an actual launch date on its interaction models. Until then, the Bloomberg appearance is brand positioning, not product news. A demo with paying customers or a named enterprise partner would change that quickly.
For founders and operators tracking the next wave of AI interface design, Murati’s 200-millisecond, continuous-stream framing is worth noting.
If it ships, it represents a meaningfully different interaction paradigm than the prompt-response loop that most current AI products are built on, a shift Relve, an AI trends intelligence platform tracks alongside the broader debate on job displacement and governance in AI.
