Separating AI hype from reality.

Noeri Announces $110M Funding to Build 'Self-Improving' Enterprise AI

November 21, 2026 · Funding

AI lab Noeri announced its launch and $110 million in initial funding on November 16, 2026. The round was led by Menlo Ventures, with participation from Madrona Venture Group, Kindred Ventures, Wing Venture Capital, and NEA. Strategic investors include Cisco Investments, NVIDIA, and Snowflake [1][3][6].

The company was founded by Stanford computer science professors Yejin Choi and Carlos Guestrin, who serve as co-CEOs. Choi is a MacArthur Fellow and two-time TIME100 Most Influential People in AI honoree (2023, 2025) whose research focuses on making smaller models more capable. Guestrin co-founded Turi, acquired by Apple in 2016, and later built and led Apple's central machine learning organization. He also co-created XGBoost, one of the most widely used open-source machine learning libraries [1][4][6].

What Noeri Claims

Noeri's pitch centers on a limitation of current AI: models are "frozen once training is done" and cannot improve after deployment [1]. The company says it builds "AI as a living system: one that self-organizes and self-improves through its interactions and experiences on the job, and grows more capable the longer it works" [1].

According to the announcement, Noeri's system studies an enterprise's data and workflows, then improves "the whole system around a task: the harness, the skills, the instructions, the tools, which model handles each step and, where it pays off, the models themselves and their underlying parameters" [1]. Every proposed change is explained and sent to a human for approval, "much like a code review" [1]. The platform works with both closed and open-weight models, and customers retain ownership of their data and the workflows and models they develop, "subject to applicable third-party model licenses" [1].

The company also says the same machinery can be applied to scientific discovery—forming hypotheses, running experiments, and learning from results on problems ranging from molecules and proteins to complex technical systems [1].

The Research Behind the Claims

Noeri says its technology draws directly from its founders' research:

  • Test-time training (Choi and Guestrin joint work): demonstrated that on-the-job learning after deployment can open new capabilities leading to novel discoveries [1].
  • Self-distillation (SDPO) (Guestrin): showed a model can become its own teacher, turning mistakes into lessons and solving hard problems in fewer attempts [1].
  • Agentic orchestration (Choi): revealed a small model conducting a team of other models and tools can outperform frontier models at a fraction of the cost [1].
  • Pluralistic alignment (Choi, as Schmidt Sciences AI2050 Senior Fellow): shown how AI can align to diverse human values and needs [1].

Unite.ai's coverage adds that Noeri's agents "learn from tasks, failures, corrections and real production traffic without requiring a large labeled dataset" and can "refine their skills, tools and orchestration, create tools when existing ones fall short, and train models on customer work" [4]. Quality and cost are measured on the customer's own traffic and metrics rather than public benchmarks [4].

What Isn't Shown Yet

The announcement provides no public demo, benchmark results, or named enterprise customers. Noeri says it is "working with strategic partners across industries, including some of its investors, deliberately choosing a varied group so its technology learns to generalize before it specializes for individual customers" [1]. Menlo Ventures partner Venky Ganesan said the firm moved quickly because "Carlos and Yejin are doing world-class research and putting it to work inside companies at the same time" [1], but no specific deployments or outcomes are described.

The announcement does not describe any production deployments where a deployed system modifies its own parameters and architecture, even with human review.

Strategic Investors

Cisco Investments, NVIDIA, and Snowflake participated as strategic investors. Their participation signals interest in Noeri's direction but does not constitute technical validation.

The Hiring Signal

Noeri is hiring researchers and engineers in Palo Alto and Zurich for "open problems in self-improving AI" [1], suggesting the core technical challenges are still being worked through. The company was founded in 2026, making this a very early-stage lab despite the large initial round.

What to Watch

The immediate test, as Unite.ai notes, is whether Noeri's partnerships "can turn its research into repeatable improvements on customers' own measures of quality and cost, while preserving the human approval and ownership controls at the center of its product" [4]. Until then, the "self-improving" label describes a research agenda, not a demonstrated capability.

Sources

  1. Noeri, Founded by Leading Stanford AI Professors, Launches with $110 ... (businesswire.com)
  2. Noeri, Founded by Leading Stanford AI Professors, Launches with $110 ... (sg.finance.yahoo.com)
  3. Noeri Launches With $110 Million to Develop Self-Improving Enterprise AI (citybiz.co)
  4. Noeri Raises $110M to Build Self-Improving AI for Enterprises (unite.ai)
  5. The Week’s 10 Biggest Funding Rounds: $400M For Neurotech Leads, Followed By Big Deals For AI, Fintech, Biotech And More (news.crunchbase.com)
  6. Noeri, Founded by Leading Stanford AI Professors, Launches with $110 Million to Empower Companies to Build and Own Their AI (businesswire.com)