Mira Murati, the former chief technology officer at OpenAI and one of the most closely watched builders in Silicon Valley, is now steering her own public-benefit corporation, Thinking Machines Lab.
Founded in February 2024, the startup has rolled out its first product, Tinker, and reached a $12 billion valuation after raising $2 billion in seed funding, positioning itself as a standard-bearer for AI accessibility and customization.
The Mira Murati Significance
A persistent friction in the AI world is the gap between cutting-edge capability and practical access. Tinker aims to collapse that gap by letting developers fine-tune leading open models, including Meta’s Llama and Alibaba’s Qwen, “with just a few lines of code,” abstracting away distributed training complexity that usually demands specialist know-how and major compute budgets. The company’s framing is plain: frontier capabilities should be usable by many, not few.
The product: Tinker
- What it is: A flexible API for fine-tuning language models, launched in October.
- What it promises: Lower the barrier to experimentation and deployment; push model customization toward mainstream developers and researchers.
- Why it’s different: It treats fine-tuning as infrastructure, repeatable, simple, and safe to adopt, instead of a bespoke engineering project each time.
The founder’s arc
Murati’s trajectory is defined by hands-on engineering and human-computer interaction before AI leadership:
- Early life and education: Raised in Vlore, Albania, schooled through a United World Colleges scholarship at Pearson College (Canada), and later completed degrees in mathematics (Colby College, 2011) and mechanical engineering (Dartmouth Thayer, 2012) .
- Early career: Interned at Goldman Sachs (Tokyo), worked at Zodiac Aerospace, then joined Tesla in 2013 as senior product manager for Model X. In 2016, she became VP of product and engineering at Leap Motion, sharpening her focus on human-computer interaction .
- OpenAI years: Joined June 2018 (applied AI & partnerships), rose to SVP (2020) and CTO (2022), briefly served as interim CEO in November 2023 during leadership turbulence, and resigned in September 2024 to “pursue my own exploration” .
- Recognition: Featured in Fortune’s 100 Most Powerful Women (No. 57, 2023), TIME’s 100 Most Influential in AI (2024), and awarded an honorary PhD by Dartmouth (June 2024) for contributions to technology and engineering.
A principled structure: public-benefit corporation
Thinking Machines Lab is organized as a public benefit corporation, a choice that aligns legally and culturally with Murati’s stated view that AI should extend individual agency and be distributed “as widely and equitably as possible.” That stance, practical, not performative,has guided early hiring and product choices, including a team of ~30 researchers and engineers drawn from OpenAI, Google, and Meta.
Growth has not been without strain. Meta CEO Mark Zuckerberg reportedly sounded out an acquisition earlier this year; when rebuffed, Meta mounted an aggressive recruitment push, dangling extraordinary packages reportedly ranging from $200 million to $1.5 billion in a bid to hire from the ~50-person startup. In mid-October, co-founder Andrew Tulloch departed for Meta amid this wider talent war. As of that report, no other employees had accepted Meta’s overtures.
The bigger shift Tinker signals
- From model worship to tooling pragmatism: Tinker leans into a future where customization beats one-size-fits-all, where the competitive edge comes from how quickly organizations make models their own.
- From power concentration to capability diffusion: Fine-tuning as a service is a pressure valve on concentration risk, potentially broadening who gets to ship high-quality AI beyond the biggest labs and platforms .
What to watch next
- Adoption beyond early adopters: Whether Tinker crosses from AI-native startups into regulated industries and traditional enterprises will test its abstraction of training complexity.
- Independence under heat: With an active poaching and acquisition climate, can the company retain talent and maintain its public-benefit mission while scaling?
- Ecosystem fit: Support for additional frontier and domain-specific models and how Tinker handles safety, evals, and governance at scale, will determine its staying power.
Mira Murati is betting that the next wave of AI adoption won’t be decided only by who trains the biggest model, but by who makes sophisticated customization dead simple. With Tinker out in the wild and a mission pinned to accessibility, Thinking Machines Lab is positioning itself as a counterweight to winner-take-all dynamics and a proving ground for how far a principled, product-first approach can go under pressure from Big Tech’s gravitational pull.
Also Read: Dutch watertech startup HULO raises €2.3m
