Noeri, a Palo Alto startup founded by Stanford professors Yejin Choi and Carlos Guestrin, has launched with $110 million in seed funding led by Menlo Ventures. The round, announced November 18, 2026, also drew Nvidia, Cisco and Snowflake alongside venture firms Madrona, Kindred Ventures, Wing Venture Capital and NEA. The money is meant to answer a problem enterprises keep running into: AI agents that make the same mistakes over and over because nothing in the system actually learns from them.
What is Noeri actually building?
Noeri's pitch is AI systems that improve themselves on a company's own work, rather than staying frozen at whatever a foundation model shipped with. Instead of a static model that answers a prompt and forgets the outcome, Noeri's software is designed to track whether its own outputs succeeded or failed inside a specific company's workflows, then adjust. The company frames this as letting businesses "build and own" their AI rather than renting a general-purpose chatbot that never gets measurably better at their specific operations.
That distinction matters because most enterprise AI deployments today plateau fast: a model answers well on day one and no better on day two hundred. Noeri is betting that self-correction, not just bigger base models, is the next real lever for enterprise AI value, and that companies will pay to own that improving system rather than lease someone else's static one.
Why does the founding team matter here?
Choi, a 2022 MacArthur Fellow, led common-sense reasoning research at the Allen Institute for AI and spent years at the University of Washington before leaving UW and Ai2 in 2024 to join Nvidia as a distinguished scientist. Guestrin co-founded Turi, the machine learning platform Apple acquired in 2016, and went on to run Apple's central machine learning and AI organization. Both are heavyweight academic names with direct product and acquisition experience, which is part of why Nvidia, Cisco and Snowflake signed on as strategic investors rather than just financial backers.
Choi started Noeri in May 2026 after leaving Nvidia; the company is based in Palo Alto with a second office in Zurich, giving it a toehold in Europe's AI research talent pool from day one.
How does this compare to other enterprise AI bets?
Noeri enters a crowded field of startups promising to make enterprise AI "actually work," from agent-orchestration platforms to fine-tuning shops. What sets the $110 million figure apart is that it arrived as a seed round, not a Series A or B, reflecting how much capital investors are now willing to commit purely on founder pedigree and a thesis, before a shipped product has proven itself at scale. Nvidia's participation also signals the chipmaker is placing bets across the application layer, not just selling hardware to whoever builds there.
What should companies watch next?
The real test for Noeri will be whether its self-improving systems produce measurable gains inside paying customers' workflows within the next year, and whether enterprises are willing to let an AI vendor's software quietly retrain itself on sensitive internal data. Expect Noeri to publish case studies or benchmark comparisons against static foundation-model deployments as it moves from launch to first contracts.