
Company Overview
Noeri is an AI lab focused on enterprise and scientific research. Rather than creating yet another general-purpose chat interface, it aims to help AI perform better and better in specific business contexts: understanding business tasks, using execution results to refine methods, and embedding those improvements into tools, processes, and models.
The company was founded by a Stanford professor of computer science Yejin Choi and Carlos Guestrin Founded with teams based in Palo Alto, California, and Zurich, Switzerland. The team in Palo Alto has long been engaged in research on common-sense reasoning, language models, and multi-value alignment; the team in Zurich co-founded Turi, a machine learning company acquired by Apple, and helped develop XGBoost. Both teams share a common goal: to connect model research to the real-world problems that businesses need to solve.About the Team
Noeri is better suited for businesses that already have well-defined AI workflows and can provide business feedback, rather than individual users who simply want to ask occasional questions or generate copy.
Milestones
Noeri made a public appearance on November 16, 2026 (local time; November 17, Beijing time), and announced that it had secured $110 million in funding, led by Menlo Ventures, with participation from Madrona, Kindred, Wing, NEA, as well as Cisco Investments, NVIDIA, and Snowflake. The announcement did not specify the funding round or valuation.
These funds are earmarked for the construction of a new laboratory and business expansion, rather than for the acceptance of the company’s full product capabilities. The company is currently working with cross-industry strategic partners to advance the application of its solutions; therefore, the descriptions on the official website should not be interpreted as standardized products that are immediately available for purchase by all companies.Company Announcement
Products & Services
Noeri's services are centered around the AI agents that enterprises already have or plan to build, with the core beingImprove the entire task system, rather than just replacing a single large model.
- Identifying Areas for Improvement Based on Business Operations: Identify problems by analyzing task results, failures, and manual corrections to determine whether the root cause lies in the instructions, tools, task assignment, or the model itself.
- Optimizing the Implementation Method: Adjust the tools and collaboration processes for the agents, and, when necessary, train specialized models better suited to specific stages.
- Measuring Actual Returns: Evaluate changes based on the company’s own traffic, quality metrics, and costs, rather than relying solely on public rankings.
For example, if an internal document review process repeatedly overlooks the same types of issues, it’s worth considering not only whether “switching to a more expensive model would yield more accurate results,” but also whether the organization of materials, verification procedures, and feedback mechanisms are appropriate. Since these tasks are repetitive, it’s easier to establish a comparison between the situation before and after improvements.
Access is available through the official website Create with Noeri Contact the team and explain the type of AI agent you wish to develop or improve. The official website does not provide a standardized, publicly available price list, and the service is not a free application that can be used immediately after registration. When discussing the project, it is advisable to prepare a specific workflow, representative samples, evaluation criteria, and a list of operations that require human approval.
How to Choose a Development Platform
| programmatic | Key Focus | Suitable Needs |
|---|---|---|
| Noeri | Enterprise-Specific AI and Its Continuous Improvement | We hope to work together to optimize the system based on our business objectives, rather than simply purchasing development components. |
| LangSmith | Engineering capabilities such as agent evaluation, monitoring, and deployment | We already have a development team and need to establish processes for testing, debugging, and deployment. |
The two are not simply interchangeable. Regardless of which approach a company takes, it must still define what constitutes success, which failures are unacceptable, and how to determine whether a change is worth deploying.
core technology
Noeri's research covers continuous learning, self-distillation, tool orchestration, and specialized model training. Put simply, it aims to enable AI not only to keep a record of its work but also to adapt its problem-solving methods based on feedback.
Listed on the official website's research page TTT-Discover, SDPO, ToolOrchestra Tasks such as continuing to learn in response to specific problems, training using structured feedback, and coordinating multiple models and tools. While these studies provide technical direction, performance on research tasks cannot be directly translated into production gains for a particular company.Research Interests
Control mechanisms are equally important. The company explains that the proposed changes would be explained to users and submitted for approval; customers retain ownership of their data and the workflows and models they build, but remain subject to the relevant third-party model licenses. Therefore, “the company owns its own AI” does not mean it can ignore the terms of use for the underlying models.
development prospect
What makes Noeri stand out is that it shifts the competition in enterprise AI from “who can deploy the most powerful model” to “who can incorporate their own business experience more quickly.” If a process is repeated every day, a single improvement can continue to yield benefits in subsequent tasks.
The challenges of this approach are also clear: feedback may be noisy, and new versions may improve one type of task while harming another. For potential clients, what’s more important to examine than the amount of funding raised is whether the modifications are explainable, whether they can be reversed, whether different tasks are evaluated independently, and whether the cost of improvement is lower than the benefits it brings. Noeri is worth keeping an eye on, but it’s better suited for discussing real-world business problems with the team rather than interpreting “self-improvement” as automated evolution that requires no management.
data statistics
Related Navigation

New York-based video creation AI company founded in 2021 with a market cap of more than $500 million

Anyscale
Valued at over $1 billion, focused on distributed computing, founded in California in 2019 year

Luchen Technology
Focusing on the research, development and application of AI big model technology, we provide efficient and low-cost AI big model solutions for various industries through our self-developed deep learning system, Colossal-AI.

WiseGen Robotics
Focusing on the research, development and production of humanoid robots and embodied intelligence products, we are committed to promoting the deep integration and innovative application of AI and robotics.

Superblocks
Focused on providing programmable platforms and AI-driven tools for enterprises, it helps developers and non-technical people to rapidly build customized internal applications, workflows and automated tasks, accelerating digital transformation and internal software innovation.

4Paradigm
Focuses on using machine learning technology to provide intelligent solutions for enterprises, helping them achieve digital transformation and business growth.

AInnovation
A company focusing on artificial intelligence technology, providing products and services with core technologies such as machine vision and natural language processing, helping the digital transformation and intelligent upgrading of several industries.

Tutor Intelligence
A robotics company spun out of MIT CSAIL combines models, hardware, remote operation, and on-site deployment to offer the Cassie and Sonny robots for palletizing, depalletizing, case packing, and piece-by-piece picking.
No comments...
