Kivorent News

Wednesday, November 18, 2026

Tech and AINews

Noeri, from Yejin Choi and Carlos Guestrin, launches with $110 million

Noeri, an AI startup led by Stanford professors Yejin Choi and Carlos Guestrin, came out of stealth on November 16 with $110 million led by Menlo Ventures. It builds AI systems that keep improving from a company's own work, with people approving each change.

Cortex, covering AI models, labs and agents
, 2 min read

$110MNoeri's first funding, led by Menlo Ventures

What to know

  • Menlo Ventures led the round; Madrona, Kindred Ventures, Wing, NEA, Cisco, Nvidia and Snowflake also invested, according to The Next Web.
  • Dealroom reports the company was founded about five months ago; it is based in Palo Alto with an office in Zurich.
  • Valuation was not disclosed, and no customers were named.

Noeri, a new AI company led by Stanford computer scientists Yejin Choi and Carlos Guestrin, launched on November 16 with $110 million in funding. Menlo Ventures led the round, The Next Web reports, with Madrona Venture Group, Kindred Ventures, Wing Venture Capital and NEA taking part, along with strategic investors Cisco Investments, Nvidia and Snowflake. The valuation was not disclosed.

Dealroom reports the company was founded about five months ago. Noeri is based in Palo Alto with an office in Zurich and is hiring in both, according to The Next Web.

The founders

Choi and Guestrin are co-founders and co-CEOs. Choi is a MacArthur Fellow who was a professor at the University of Washington and a senior director at the Allen Institute for AI, and has also worked at Nvidia, according to The Next Web and Securities.io. Guestrin co-founded Turi, which Apple bought in 2016, then led Apple's machine learning and AI organization; he co-created the XGBoost library and directs the Stanford AI Lab.

What Noeri sells

Noeri describes its product as self-improving AI for businesses and scientific research. Instead of one general model, it tunes the parts of a company's AI setup: its instructions, its tools, which model handles each step and, where it helps, the models themselves. Each proposed change comes with an explanation, and a person approves it, much like a code review, The Next Web reports.

Securities.io describes a "production loop" that turns each task, failure and correction into a lesson, so an agent improves from its own traffic and is measured on the customer's own metrics rather than public benchmarks. Noeri works with both closed and open-weight models, and customers keep ownership of their data and of the workflows and models they build.

On the science side, the company says smaller, specialized models can form hypotheses and run experiments on problems involving molecules and proteins at lower cost. Dealroom sums up the pitch as many smaller, purpose-built systems rather than one large one. Guestrin said "every company we talk to is riding the same curve," per The Next Web.

No commercial customers were named. Securities.io reports unnamed proof-of-concept partnerships in enterprise and science. Pricing and general availability were not announced.

A busy month for agent tooling

The raise comes days after LangChain raised $300 million for its agent platform, and as large labs push their own agent products, such as Anthropic's self-hosted Claude Managed Agents. Nvidia, which invested, continues to back AI startups after Jensen Huang's resignation as CEO.

Why it matters

Many companies already run AI agents but find they stall after launch. Noeri is betting that a system which keeps tuning a company's own setup, with a human signing off on each change, can beat simply swapping in a bigger model. With $110 million and two well-known researchers, it starts with more money than most new AI labs, but it has yet to name a customer.

Sources

AI agentsNvidiaAI models

Written by an AI correspondent under standards set by editor Jacob Knight. Spotted an error? Tell us and we will fix it.