Noeri, a Palo Alto AI lab founded this year by Stanford professors Yejin Choi and Carlos Guestrin, came out of stealth on Monday, Nov. 16, with $110 million in funding. Menlo Ventures led the round, with Madrona Venture Group, Kindred Ventures, Wing Venture Capital and NEA participating, and Cisco Investments, Nvidia and Snowflake joined as strategic investors, according to The Next Web. AI Weekly, summarizing Bloomberg's report, describes the round as a seed and the company as 5 months old.
Noeri did not disclose a valuation, Bloomberg reported. The company is headquartered in Palo Alto with a second office in Zurich, and its pitch goes straight at a problem most AI startups in the South Bay are dealing with right now: model bills that keep climbing as usage grows. Below is what Noeri builds, who is behind it and what it could mean if you run AI in production at your own company.
What Noeri builds: AI that keeps improving on the job
Noeri says most AI models stop learning once training ends, while its systems keep improving as they work. Its AI studies a company's own data and workflows, then changes the system around a task, including the instructions, the tools, which model handles each step and, where it pays off, the models themselves. Pulse 2.0 adds that the system also tunes skills and model selection, and that the approach draws on the founders' research in test-time training, self-distillation, agent orchestration and alignment.
The detail founders will like most is the review step. Each proposed change comes with an explanation, and a person approves it, much like a code review. The system works with both closed and open-weight models, and customers keep ownership of their data and of the workflows and models they build, subject to third-party licenses.
Who founded Noeri: Yejin Choi and Carlos Guestrin
Both founders serve as co-CEOs. Choi is a MacArthur Fellow and a former University of Washington professor and senior director at the Allen Institute for AI, and Bloomberg reported that she was also a distinguished scientist at Nvidia. Guestrin co-founded Turi, which Apple bought in 2016, and then led Apple's central machine learning and AI organization. He also co-created XGBoost, the open-source machine learning library, and Bloomberg describes him as director of the Stanford AI Lab.
Choi told Bloomberg the idea came partly from work with a Stanford microbiology and immunology team that wanted cheaper ways to use OpenAI's models. Her group built a small AI that learned to use other models, and she said it beat GPT-5 on that team's research needs. Guestrin told The Next Web that every company he talks to faces the same trade-off between paying more for the best model and accepting worse results from a cheaper one, and that specialization can improve both sides of that curve.
Why Noeri matters for founders running AI in production
If you're building an AI product in Palo Alto or anywhere else on the Peninsula, you've probably lived through that trade-off. Frontier model costs grow with every new customer, and swapping in a cheaper model can break quality in ways you only discover after users complain. Noeri's bet is that many smaller, specialized systems, tuned on your own data and checked by a person, can improve cost and quality together.
The investor list is also a signal. Nvidia, Snowflake and Cisco joining as strategic investors suggests the chip, data and networking companies want to stay close to whoever decides how enterprise AI work gets routed. For operators, the review-first design answers a real fear about systems that modify themselves, since nothing changes in production until a person signs off. Guestrin also told Bloomberg that he and Choi worry about one company owning everything, which helps explain why customer ownership of models sits at the center of the pitch.
Noeri is entering a crowded field. This week we wrote about LangChain's Series C and the agent tooling race, and Corma's $125 million Series A showed how quickly specialized AI labs can raise this fall. If you're weighing a Stanford-connected path for your own company, our guide to the StartX accelerator explains how the Stanford-affiliated program works. On the robotics side, Genesis AI raised a $500 million Series A the day after Noeri launched, with offices in San Carlos.
Noeri is hiring in Palo Alto and Zurich
Noeri says it is hiring researchers and engineers in both Palo Alto and Zurich, and it is working with partners across industries, including some of its investors. The company also wants its AI to work with scientists, forming hypotheses, running experiments and learning from the results on problems ranging from new molecules and proteins to complex technical systems. It says smaller, specialized models make that kind of search affordable.
For machine learning engineers in the South Bay who want research depth without a commute to San Francisco, a well-funded lab led by two working professors is an unusual option. Expect competition for those roles from people who already know the founders' published work.
Meet AI builders in Palo Alto and Redwood City
Palo Alto has been home to a lot of Tribe's AI programming. In September, Tribe hosted an AI Infra Dinner in Palo Alto with Prateek Joshi, a small table of 12 to 15 seats built for deep conversation about the same stack Noeri wants to tune. Gatherings like that one are part of South Bay Renaissance, which Tribe organizes to rebuild the startup and social community of the South Bay and Peninsula in person, with founder dinners, AI deep dives, demo nights and talks.
The next stop on the Tribe Events calendar on Luma is the South Bay Renaissance Product Builders Mixer in Redwood City on Thursday, Nov. 19. Follow the calendar to catch upcoming AI dinners and talks as soon as they're posted.
Talk about new AI labs with people who follow them
Funding announcements like this one are more fun to pick apart with people who track AI closely. Join the AI Trends Tribe to trade notes on new labs, model releases and what they mean for the products you're building, and keep the Tribe Events calendar handy so you can meet the same crowd in person.