nao Labs: Product, team and how it was built
nao Labs (Open-source analytics agent for data teams, France): 2.0K GitHub stars, November 2026. Evidence grade E: no revenue figure was checked. Team: 2 founders; Y Combinator lists a team of 2. Launched 14 May 2025; still running.Added
- Outcome
- 2.0K GitHub stars, November 2026
- Evidence
- E no revenue figure was checked
- Launch
- 14 May 2025
- Status
- Still running
- Team
- 2 founders; Y Combinator lists a team of 2
- Outside funding
- Y Combinator Spring 2025 batch; the amount is not stated in the sources read
- Main channel
- Not stated
What is not verified
No revenue figure was found, so the story is grade E. The company, pricing, terms, privacy and blog pages, the Y Combinator pages, GitHub's API and the documentation were read, and none gives revenue. A reader could look for a payment-processor profile or a press interview. No source states that a customer has paid, and the funding amount was not found. The GitHub license field reads Other, but the repository's LICENSE file is the Apache License 2.0 text. This story is grade E: the business is real and its product, price and customers are described from public pages, but no revenue figure was checked. Aiolum shows it as an idea worth studying, not as proof that it makes money.
nao is an analytics agent. It answers business questions by running SQL, the query language of databases, on the company's data warehouse. The data team builds the context layer, a set of files with table descriptions, metric definitions and rules that the agent reads. The core is open source under the Apache 2.0 license, and nao Labs also sells a hosted version, nao Cloud.
Y Combinator lists Claire Gouze and Christophe Blefari as founders, the year 2024, the Spring 2025 batch and a team of 2. An Internet Archive copy of 1 September 2024 shows a no-code data tool in beta. The first launched product was an AI code editor for data work, in a Y Combinator post of 14 May 2025. The company blog announced the open-source analytics agent on 10 February 2026.
No revenue figure was found, so the story is grade E. The strongest checked fact is 2,004 GitHub stars on 12 November 2026, read from GitHub's public API. A self-service paid plan exists, with a Stripe subscription path in the documentation. No source says how many customers pay.

Why did nao Labs work?
Four causes, in the order they mattered. Each one is a reusable lesson.
The founders came from data work and built for data teams
Y Combinator's launch post says Christophe and Claire each spent 10 years in data. Their profiles list a head of data role and work on data tools and platforms at more than 10 companies. The launch post says data teams cannot keep the business pace. Their first product served those teams, and the later product does too.
The product moved from an editor to an open-source agent
The first launched product, from May 2025, was a code editor for SQL and dbt work. In February 2026 the company blog said it was open sourcing nao as an analytics agent framework. It wrote that closed products hide how an agent reaches an answer. The code on GitHub lets a data team read and change the agent.
It charges for hosting while the core stays free
The pricing page says the core is open source and that nao Cloud is the same product, hosted and run by the company in Europe. It lists nao Cloud as free for 3 users and US$50 per user a month after that. The pricing page lists “Bring your own LLM key, no token margin”. The billing documentation also sells nao credits for managed models, with “A 5% nao platform fee” on each purchase.
How it was built
These details come from the company's own pages and from GitHub. They are grade C, and the language is grade B. [7, 6, 14, 12]
- The repository's main language is TypeScript. The command line tool nao-core is installed with pip, and the chat interface is also published as a Docker image.
- Context is kept as files in a Git repository: one folder per table, plus rules, documents and dbt files. The llms.txt page says the agent searches these files directly.
- A command called nao test runs question-to-SQL tests and compares the data returned. The llms.txt page says this is a data comparison and not a judgment by a model.
- The privacy policy says nao Cloud runs on Google Cloud Platform in the European Union and that usage events go to PostHog. The billing documentation names Stripe.
- The privacy policy says nao does not provide a language model and the customer uses its own API key. The billing documentation, however, describes managed models paid with nao credits.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before nao Labs, and why did they fail?
Two data-tool companies in our database of failed startups failed after promising to make data work easier. One sold an add-in for analysts, and one was an open-source database. The values below are from that database and are not independently verified.
DataNitro
The promiseRun Python scripts inside Excel, so that data analysts could move spreadsheet work to Python.
Why it diedThe record says the offer was limited by Excel's architecture. It adds that Microsoft Excel has since added Python itself.
RethinkDB
The promiseAn open-source database that pushes changed query results to applications, so that developers build real-time features easily.
Why it diedThe record says developers liked it but would not pay for it. It adds that an open-source product left only enterprise support and hosting to sell.
The failed startups come from our database of 1,749 dead startups, retold here in our own words. Its entries are AI-assisted summaries of public sources, so these cards carry no evidence grade.
What changed
DataNitro depended on one vendor's product, and the vendor added the feature. The nao core is open source. The llms.txt page lists several warehouses and model providers that it works with.
RethinkDB could sell only support and hosting around its open-source code, the record says. nao Labs sells hosting around an open-source core too. No source shows that this earns enough. The checked evidence is use of the code, not revenue.
Can one person copy nao Labs now?
The open-core model, yes. The revenue, not yet shown.
A small team with long experience in one field can release the core of a product for free and charge for hosting. The sources show a price list and a payment path. They show no revenue, so a reader should treat this as a pattern and not as proof that it earns.
Where it is still open: Whether the cloud plan earns enough for a company of 2, and how many teams pay for it.
- 1Pick a field where you worked for years and know the daily problems.
- 2Release the core as open source and say in public why.
- 3Sell the same product hosted, free for a small team and per user above that.
- 4Let customers bring their own model key, so that you carry no model cost.
How much revenue does nao Labs make?
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Company started | 2024 | C | The company page says it started nao in 2024, and Y Combinator lists the year 2024. The earliest Internet Archive copy of the site is from 1 September 2024. It shows the same name and a no-code data tool in beta. No earlier name was found. | [1, 2, 16] |
| Founders | Claire Gouze (CEO) and Christophe Blefari (CTO) | C | Y Combinator's company page and the company's llms.txt name them. The company page says Claire was head of data at sunday and Christophe built data stacks at three companies. | [1, 2, 6] |
| Earlier product | AI code editor for data, launched 14 May 2025 | C | Y Combinator's launch post is titled nao Labs - Cursor for data and was created on 14 May 2025. An Internet Archive copy of the home page, from 8 August 2025, reads AI code editor for data. It has a download button and a Pricing menu. Before it, a copy from 1 September 2024 offered a no-code data tool with a Join beta button. | [3, 8, 16] |
| Change of direction | open-source analytics agent, announced 10 February 2026 | C | The company blog post of 10 February 2026 says it is open sourcing nao as an analytics agent framework. A post of 12 February 2026 says it released its first fully open source product. | [9, 10] |
| GitHub stars, 12 November 2026 | 2,004 | B | Read from GitHub's public API for the repository getnao/nao, which was created on 30 December 2025. The main language is TypeScript. | [7] |
| First GitHub release | v0.0.45, 12 February 2026 | B | The oldest entry of the repository's release list, which holds 112 releases. The newest, v0.4.2, is dated 10 November 2026. Older version tags exist without a release entry. | [7] |
| Team | 2 people, Paris, France | C | Y Combinator lists a team size of 2 and the location Paris, France. The company page says it is building between Paris and San Francisco. An Internet Archive copy from 1 September 2024 lists three team members, with a chief operating officer who is not on today's pages. | [1, 2, 16] |
| Legal entity | nao Labs Inc., a Delaware corporation | C | The terms and conditions, effective 12 October 2026, name this entity as the party to paid subscriptions. | [11] |
| Funding | Y Combinator Spring 2025 | C | The home page says Backed by Y Combinator. No source read states an amount. | [1, 4] |
| Self-service paid plan | 14-day free trial without a card, then a Stripe subscription | C | The billing documentation says admins subscribe in Settings through Stripe. The terms accept a subscription completed through a payment link. | [12, 11] |
| Cloud price on the pricing page | US$50 per user a month, free up to 3 users | C | The pricing page lists a Get started for free button that points to the sign-up page. | [5] |
| Time from launch to the stars count | 18 months | calc | Calculated: 14 May 2025 to 12 November 2026. | [3, 7] |
How to read this table. Grade B is a value read directly from a platform. Grade C is a statement by the company on its own pages. Grade X is calculated on this page. No value is a revenue figure.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Open-source | 0 | 0 | self-hosted, unlimited users, Apache 2.0 |
| Cloud | 45 | 50 | free up to 3 users, then per user; capped at 2,000 a month or 20,000 a year |
| Enterprise | custom annual per-seat license; contact the company |
The open-source version is free and self-hosted. The pricing page lists nao Cloud as free for up to 3 users, then US$50 per user a month. It caps this at US$2,000 a month, with unlimited users from 40. Billed yearly, it lists US$45 per user a month, capped at US$20,000 a year. The billing documentation names a plan called Unlimited at US$2,000 a month. Customers can bring their own LLM key and pay the model provider. The billing documentation also offers nao credits for managed models, with a 5% platform fee on purchases.
When did nao Labs launch, and how did it grow?
From a 2024 start and an AI code editor in 2025 to an open-source analytics agent and a cloud plan in 2026. Each entry carries its grade.
- 2024CThe company page says the founders started nao in 2024. Y Combinator lists the year 2024 as the founding year. On 1 September 2024 the site offered a no-code data tool in beta. [2, 1, 16]
- 14 May 2025CY Combinator launch post for nao, an AI code editor that works with the user's data warehouse. The founders ask readers to download it. [3]
- 10 February 2026CThe company blog announces that nao becomes an open-source analytics agent framework. The repository had been created on 30 December 2025. [9, 7]
- 12 February 2026BFirst entry in the GitHub release list, v0.0.45. The company blog describes its first fully open source product the same day. [7, 10]
- 29 October 2026CThe blog announces that anyone can sign up on nao Cloud, connect a warehouse and deploy the agent without the command line. A free trial starts with the first project. [13]
- 12 November 2026B2,004 GitHub stars. The pricing page lists a free open-source version, nao Cloud at US$50 per user a month and a custom Enterprise plan. [7, 5]
What could go wrong for nao Labs?
The risks are the mirror image of the causes above.
What is not verified
No revenue figure was found, so the story is grade E. The company, pricing, terms, privacy and blog pages, the Y Combinator pages, GitHub's API and the documentation were read, and none gives revenue. A reader could look for a payment-processor profile or a press interview. No source states that a customer has paid, and the funding amount was not found. The GitHub license field reads Other, but the repository's LICENSE file is the Apache License 2.0 text.
Home page badges and counters
The home page says Backed by Y Combinator, which the Y Combinator directory confirms. It also says The #1 open source agentic BI builder, with no ranking source. It shows a SOC 2 badge, and the Trust Center page was not read. The counters 6,310, 24,184 and 19 sit inside a demo screen, and the stars, forks and contributors counters are empty placeholders.
Four customer testimonials come from the company's own pages
The home page and the users page show four quotations from named data leaders at LiveKit, Finfrog, Qare and Everyday. The users page invites readers to join its wall of fame. No outside source confirms them. The same quotation appearing elsewhere under another name could not be checked, because the search service was unavailable.
Pricing wording differs between pages
The pricing page says “Bring your own LLM key, no token margin”. The billing documentation says of credit purchases for managed models: “A 5% nao platform fee and applicable taxes are added”. The documentation also names a per-seat plan and an Unlimited plan at US$2,000 a month. The terms name nao Labs Inc., a Delaware corporation, while Y Combinator lists the location as Paris, France.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- C Y Combinator company page: nao Labs, read 12 November 2026Read in full. Source for the founders, the year 2024, the Spring 2025 batch, the team size of 2 and the location.
- C nao company page: We turn data teams into context engineers, read 12 November 2026Read in full. Source for the start in 2024, the founders' backgrounds and the offices in Paris and San Francisco.
- C Y Combinator launch post: nao Labs - Cursor for data, 14 May 2025Read in full as the page's data. Source for the first product, its date and the founders' ten years in data.
- C nao home page, read 12 November 2026Read in full. Source for the badges, counters and testimonials, and for the open-source and SOC 2 statements.
- C nao pricing page, read 12 November 2026Read in full, including the questions in its structured data. Source for the plans and prices.
- C nao llms.txt, read 12 November 2026Read in full. The company's own summary: founders, plans, canonical facts and what the product does.
- B GitHub public API: repository getnao/nao and its releases, read 12 November 2026Read the repository record and the release list. Source for the stars, the creation date, the language and the first release.
- C Internet Archive copy of the nao home page, 8 August 2025Read in full. Shows the earlier product, an AI code editor for data, with a download button and a Pricing menu.
- C nao blog: We're launching the first Open Source Analytics Agent Builder, 10 February 2026Read in full. Source for the move to an open-source analytics agent.
- C nao blog: Why we're making our Analytics Agent open source, 12 February 2026Read in full. Source for the first fully open source product and the reasons given.
- C nao Terms and Conditions, effective 12 October 2026Read in full. Source for the legal entity, payment links and Stripe as a payment method.
- C nao documentation: Billing and creditsRead in full. Source for the trial, the plans, the Stripe subscription path and the credit fee.
- C nao blog: Launching nao agentic BI builder, 29 October 2026Read the opening and the first steps. Source for the cloud sign-up without the command line.
- C nao Privacy Policy, last updated 15 October 2026Read in full. Source for hosting, PostHog, the model provider rule and the open-source telemetry.
- C nao users page: user references, read 12 November 2026Read in full. Source for the four named testimonials.
- C Internet Archive copy of the nao home page, 1 September 2024Read in full. The earliest copy the archive returned. Shows a no-code data tool in beta and three team members.
- B GitHub repository getnao/nao: LICENSE file, read 12 November 2026Read in full. The file is the Apache License 2.0 text, and its first line reads Apache-2.0.
- C Database of failed startups, local exportRecord pages read on the website on 12 November 2026. AI-written summaries, not independently verified. Source for the failed predecessors.
Run this company? Send corrections or a verified figure to aiolum@livyden.com