HomeAINoeri Launches With $110 Million to Build Company-Owned AI

Noeri Launches With $110 Million to Build Company-Owned AI

Noeri, a new company founded by the Stanford computer-science professors Yejin Choi and Carlos Guestrin, has launched with a pitch that companies should build specialised AIs from their own data and keep them. Its website carries a news item reading “Noeri raises $110 million to advance the future of plural intelligence.” The same page’s “Partners and investors” strip names eight backers: Menlo Ventures, Madrona, Kindred Ventures, Wing, NEA, Cisco, NVIDIA and Snowflake.

GeekWire reported that Choi and Guestrin are co-chief executives, that Menlo Ventures led the round, and that a valuation was not disclosed. Bloomberg first reported the funding, GeekWire said.

What Noeri says the product is

On its website, Noeri says that with the product “your company builds specialized AIs from its own data and know-how.” Those systems, it says, “keep learning from real work, team up on your hardest problems, and remain entirely yours.” The same site describes a “self-improvement loop” that, it says, works in production, learns from outcomes, upgrades tools and models, and finds gains and opportunities. It says the system “learns on the job,” so that “every task, failure and correction becomes a lesson,” and that it can refine and create tools, train its own models on a company’s work, and measure quality and cost on the company’s own traffic rather than on a public benchmark.

GeekWire’s report adds a further company description: that the software studies a company’s data and workflows, then proposes changes to the instructions, the tools, which model handles each step and, in some cases, the models themselves. Each change, the company told GeekWire, is explained and sent to a person for approval, “much like a code review.” Customers, it said, keep ownership of the workflows and models they develop, subject to third-party model licences. Those product claims are the company’s own, and they have not yet been shown in customer use.

The firm lists offices in Palo Alto and Zurich and links to a careers page. A vision essay posted with the launch argues against “a single superintelligence, owned by a few.” GeekWire quoted Choi and Guestrin writing that “few companies would build their core operations on a single consultant who also advises every competitor, can change its terms at will, and can never be brought in-house.”

The founders, and the money

Noeri’s site presents Choi as a Stanford professor of computer science and a senior fellow at the Institute for Human-Centered Artificial Intelligence, and notes a MacArthur fellowship, TIME100 recognition, and earlier research roles at the Allen Institute for AI and at NVIDIA. It presents Guestrin as a Stanford professor, director of the Stanford Artificial Intelligence Laboratory, and a member of the National Academy of Engineering, and says he co-created XGBoost, co-founded Turi, which Apple acquired in 2016, and later led machine learning and AI at Apple. Noeri, the site says, is his seventh startup.

The eight names in the site’s investor strip are Menlo Ventures, Madrona, Kindred Ventures, Wing, NEA, Cisco, NVIDIA and Snowflake. GeekWire reported that Menlo Ventures led the $110 million round, and that the valuation was not disclosed. GeekWire did not describe the round as a Series A or Series B.

The raise sits in a busy period for AI-developer and infrastructure funding already covered by TSN. LangChain’s $300 million Series C is a recent example of a large round behind software that companies use to build agents. Fractile’s $720 million Series C is a much larger, hardware-side raise in the same week of funding stories. A closer product neighbour than either is IBM and Microsoft’s Imagine Enterprise agent platform, another pitch about companies running specialised agents on their own work.

A launch and a raise are not a proven product

Noeri has put a company, a founding team and a $110 million raise on the public record. The figure and the eight investor names come from the company’s own site. What the site does not say is who led the round, what the company is worth, or how the approval loop is meant to work in practice. Those details come from GeekWire, which also said Bloomberg first reported the funding and that the valuation was not disclosed.

A well-known founding pair and a nine-figure launch can be read as if the software were already doing what the website describes. It is not yet shown to be doing so. The self-improvement loop, the claim that systems “remain entirely yours,” the human approval of proposed changes, and the argument against a single, centralised model are part of the pitch. They will be testable once named customers appear and the product is described in their own words, including what was approved, what was owned, and what improved. Ownership “subject to third-party model licences,” as the company told GeekWire, is also a reminder that the pitch still sits on top of models other people train.

The next facts to watch are ordinary ones: whether the company names customers, and whether later reporting can describe the product in use rather than only on noeri.ai. Until then, the news is that Noeri has launched and has said it has raised $110 million.

Sources

Related stories

Follow @tsnmediahub on X for breaking AI news as it happens.

Follow on X

Share this story

Latest stories

  1. TensorWave and OpenAI Agree a 457MW AMD MI455X Deal
    AI Chips & Compute
  2. OpenAI Evaluator Pilots Start With Apollo and METR
    AI Policy & Safety

More in this category

Latest stories

  1. TensorWave and OpenAI Agree a 457MW AMD MI455X Deal
    AI Chips & Compute
  2. OpenAI Evaluator Pilots Start With Apollo and METR
    AI Policy & Safety

Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.