Notte logo
Browser Automation · Web Scraping

Notte

Notte is a browser infrastructure platform for AI agents. It bundles cloud browser sessions, natural-language browser agents and serverless browser functions behind one API, with credential vaults, real agent identities, proxies and CAPTCHA handling for production web automation.

Active Free plan Freemium API available 13+ Verified by Guidaio
Overview

What is Notte?

Notte is a browser infrastructure platform for AI agents, built by Notte Labs, Inc. for teams whose software has to operate a real website rather than call an API. It positions itself as the layer between an agent and the open web, and it is organised around three pillars.

Browser Sessions are isolated cloud Chromium instances on a global edge network. The site advertises more than a thousand concurrent instances, latency under 50 milliseconds and 99.9% uptime, with autoscaling, residential proxies, antibot detection and CAPTCHA autosolving.

Browser Agents take a task described in plain language and carry it out with no selectors to write or maintain. They are model-agnostic, working with OpenAI, Anthropic and Gemini among others, and the site claims a success rate above 90%.

Browser Functions are described as AWS Lambda for browser work: serverless code running next to the browser, with a claimed zero network hop, cold starts under 200 milliseconds and cron scheduling.

Around these sits the authentication layer, which is the platform's clearest differentiator. Agent Vaults hold credentials encrypted with AES-256, scoped per session and rotated automatically; when an agent meets a login form it emits a placeholder and the platform substitutes the real secret at the browser-action layer, so the credential never enters the model's context window. Agent Identities go further, provisioning genuine email inboxes and SMS numbers per agent so a workflow can create an account and clear verification codes. Session Profiles save full browser state, letting a run log in once and skip authentication thereafter.

Existing code is welcome: Playwright, Puppeteer, Selenium, browser-use and Stagehand all connect over the Chrome DevTools Protocol, in Python, TypeScript or Node.js, and an MCP server exposes the platform to coding agents such as Claude Code and Cursor. A spin-off called Anything API turns a described web action into a bespoke endpoint returning structured JSON.

The company is a Delaware corporation from the Y Combinator Summer 2025 batch. It is SOC 2 Type II certified, but publishes neither a postal address nor any GDPR framework.

What it does

  • Launch thousands of concurrent cloud browsers around the world
  • Run browser agents from a plain-language task description, with no selectors to write
  • Deploy browser scripts as invocable API endpoints on a cron schedule
  • Let agents log in with vaulted credentials the model never sees
  • Give agents real email inboxes and phone numbers to pass one-time codes and two-factor checks
  • Extract structured data from dynamic and authenticated pages
  • Debug runs through live view, session replays and observability
Audience

When to use Notte / When not to

A quick filter to help you decide if Notte is the right fit.

When to use Notte

  • Engineering teams running browser automation in production at scale
  • Developers building AI agents that need a reliable web execution layer
  • Teams whose workflows have to log in, sign up or clear a two-factor check
  • Playwright, Puppeteer or Selenium users who want managed infrastructure without a rewrite
  • High-volume workloads where the per-browser-hour rate becomes a real line item

When not to use Notte

  • Non-developers: there is no no-code product, everything runs through an SDK, a CLI or the API
  • Organisations that require European data residency, since processing is United States only
  • Buyers who need a GDPR framework, a Data Processing Agreement or a named subprocessor list before signing
  • Anyone expecting a mobile application or an interface in a language other than English
  • One-off manual scraping, where a browser extension would be faster and cheaper
Get started

How to use Notte

A typical end-to-end flow, from setup to results.

  1. Create an account on the Notte console: no credit card, and 10 US dollars of starter credits
  2. Install the Python or TypeScript SDK and authenticate with your API key
  3. Open a cloud browser session, which the homepage example does in a handful of lines
  4. Choose a control mode: Notte actions, an AI agent given a plain-language goal, or your existing Playwright or Puppeteer code over CDP
  5. Inspect the live page with an observe call or the CLI instead of hand-writing selectors, as the documentation insists
  6. Attach a browser profile, route through residential proxies and enable CAPTCHA solving as the target site demands
  7. Store any credentials in a vault, and create a persona when the workflow has to sign up or clear a verification code
  8. Export the generated workflow code, or build it conversationally with Agent Builder or by recording it in Demonstrate Mode
  9. Deploy the script as a Browser Function to obtain an API endpoint, adding a cron schedule if it has to recur
  10. Review each run through live view or session replay, and track consumption against your plan credits
Quick read

Pros & Cons

Pros

  • One integrated stack instead of assembling a browser host, an agent framework, a credential store, proxies and a deployment target
  • Credentials are substituted at the browser layer and never enter the model's context window
  • Agent personas provision real inboxes and phone numbers, covering sign-up and two-factor flows that competing platforms leave to the customer
  • Browser time at 0.05 US dollars per hour, against the 0.10 to 0.12 dollars Notte's own comparison page attributes to Browserbase
  • A genuinely usable free tier: 10 US dollars of non-expiring credits, five concurrent browsers, no credit card
  • Vaults, replays and CAPTCHA solving available on every tier rather than gated behind a paid plan
  • SOC 2 Type II certification, a Trust Center listing 71 controls, a public status page and an open-source SDK

Cons

  • No postal address anywhere: the privacy policy invites contact via a company address it never prints
  • No GDPR framework at all, no Data Processing Agreement and no Article 27 representative, which blocks many European buyers
  • Data is processed in the United States only, with no regional residency option below the Enterprise tier
  • No subprocessor list: the privacy policy names categories of providers, never the providers themselves
  • No stated position on whether customer data is used to train models
  • Monthly plan credits expire each cycle, subscriptions renew automatically and terminating brings no refund
  • A very young company whose legal pages sit behind a bot challenge that keeps ordinary readers out
Pricing

Pricing & Plans

Notte offers a permanent free plan at 0 US dollars, granting 10 US dollars of credits once, without expiry and without a credit card. The cheapest paid plan is Developer at 20 US dollars per month, which includes 20 US dollars of credits refreshed each billing cycle. Pricing is credit-based throughout: credits are consumed by browser sessions and browser functions at 0.05 US dollars per hour, residential proxies at 10 US dollars per gigabyte, and LLM tokens at token price with no Notte markup. No annual billing option is published, so no discounted yearly rate applies.

Plan 1
  • Free — 0 USD — 10 USD of credits granted once and never expiring
  • 5 concurrent browsers
  • 5 concurrent agents or functions
  • 5 vaults
  • 5 personas
  • 15-minute maximum session
Plan 3
  • Startup — 100 USD per month
  • marked Most Popular — 100 USD of credits per month
  • 100 concurrent browsers
  • 100 concurrent agents or functions
  • 200 vaults
  • 200 personas
  • 300-minute maximum session
  • bring your own keys and proxies
Plan 4
  • Enterprise — custom pricing — custom credits and concurrency
  • bring your own keys
  • proxies and cloud
  • SSO
  • SLAs
  • HIPAA
  • targeted stealth and on-premise deployment
Plan 5
  • Usage rates across all plans — browser sessions 0.05 USD per hour
  • browser functions 0.05 USD per hour
  • residential proxies 10 USD per gigabyte
  • LLM tokens at token price with no markup
Special offers — 10 US dollars of starter credits granted once on sign-up, with no expiry and no credit card required · A 'Tell your CTO about Notte' form on the homepage which, according to the pre-written message it sends, also grants starter credits · No student, non-profit, unemployed or annual-commitment discount is published
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Notte handles your data.

GDPR overview

There is no GDPR implementation to report. The words GDPR and General Data Protection Regulation appear nowhere on the site, in the privacy policy or in the terms of use. This is silence rather than a refusal: Notte never claims compliance and never denies it. What the policy does provide is a transparency section under the EU AI Act (Regulation (EU) 2024/1689), classing the service as a general-purpose automation tool outside Annex III, and a cookie-consent gate for visitors in the European Economic Area and the United Kingdom. The named data rights are American: CCPA for California residents and a sale opt-out for Nevada. Generic access, portability and deletion rights are offered depending on jurisdiction. No Article 27 EU representative is designated, no Data Processing Agreement is offered, and no standard contractual clauses are mentioned.

Who owns the data?

The terms contain no explicit ownership clause over customer data, but they do claim a very broad licence over anything you post. By providing User Contributions you grant Notte Labs a comprehensive, perpetual, irrevocable, worldwide, transferable and royalty-free licence to use, copy, modify, create derivative works from, distribute, display, store, transmit and commercialise that material, including your name, company name, trademarks and logos. You warrant that you hold those rights and remain fully responsible for the content. The platform itself stays owned by Notte, its licensors and its providers. Users may request access to, portability of, or deletion of their personal information, subject to legal exceptions.

Reuse rights

The terms let you use the platform commercially: entities may build their own products and services on top of it, provided they do not copy, clone or substantially replicate the platform or its functionality. Data your agents extract is yours to reuse without asking permission, the terms setting no restriction on it beyond compliance with applicable law and the platform's intended use cases. Notte's own site material is treated differently: you may not reproduce, redistribute or republish it beyond browser caching and a single printed copy. On its side, Notte processes account details, billing information, usage telemetry and browser session data — pages visited, page content, URLs, screenshots and interaction logs — to run your automation, provide session replays, secure the service, analyse usage, develop features and meet legal obligations. That data is shared with unnamed infrastructure, analytics, payment, support and security providers, and may be transferred in a merger or asset sale. No position is stated anywhere on whether customer data is used to train AI models.

Data retention & training

Retention summary
Notte publishes retention rules but never a duration. Browser session data, meaning the pages an agent visits, their content, URLs, screenshots and interaction logs, is kept for the retention period attached to your plan and then deleted, but that period is stated nowhere on the public site. Personal information is held for as long as needed to deliver the service, meet legal and regulatory obligations, resolve disputes and enforce agreements, the policy citing sensitivity and misuse risk among its criteria. Once retention is no longer justified, data is described as securely deleted or anonymised. Users may request deletion of their personal information, subject to legal exceptions. Anyone needing a committed retention window will have to ask for it directly.
Trains on customer data
Unclear
GDPR contact

Hosting summary

Notte Labs is a United States company and states plainly that it processes information within the United States, where its servers and central database operate. International users are told their information may be transferred to, stored in and processed in the United States, and that using the platform constitutes consent to that transfer, including to jurisdictions whose data-protection standards differ from their own. No European, United Kingdom or Asia-Pacific hosting region is offered, and no data-residency choice appears anywhere in the plans. The only exception sits at the Enterprise tier, which advertises on-premise deployment and bring-your-own-cloud. Technically, the domain resolves to an address geolocated in the United States on Amazon infrastructure. The Trust Center documents controls covering host hardening, infrastructure as code, penetration testing and the encryption of file systems, data in transit and file stores, but it names no hosting provider and publishes no subprocessor list.

Hosting countries
🇺🇸 United States
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Notte.

  • The tool exists to log in, sign up, solve CAPTCHAs and route through residential proxies, capabilities that can breach the terms of the sites you automate, and the responsibility for that sits entirely with you
  • Storing production credentials in a third party's vault concentrates risk: the encryption and rotation are documented, but a compromise of the vault would be a compromise of every account it holds
  • Agent identities with real inboxes and phone numbers make it trivial to create accounts at scale, which shades quickly into activity many platforms forbid and some jurisdictions penalise
  • No GDPR framework, no Data Processing Agreement and United States-only processing: routing European personal data through Notte may put you, rather than Notte, in breach
  • Nothing tells you whether your session content trains models, and session data includes page content and screenshots of everything your agent sees, potentially including other people's personal data
  • The perpetual, irrevocable and commercialisable licence the terms claim over User Contributions is wider than the service requires, and extends to your name and logo
  • An agent with a success rate above 90% still fails roughly one run in ten, often silently: automating anything financial or irreversible without human review is a real hazard
Setup

Setup & Integrations

Technical difficulty

Developer-level. There is no no-code path: everything runs through the Python or TypeScript SDK, the CLI, the API or the Studio IDE. Getting a first session open is genuinely quick, the homepage showing a seven-line snippet and claiming three lines of setup, and existing Playwright, Puppeteer or Selenium code connects over CDP without a rewrite. Reliability takes longer: the documentation explicitly forbids hand-writing selectors and prescribes inspecting the live page first, then generating and editing workflow code. Agent Builder and Demonstrate Mode shorten that considerably. No credit card is needed to begin.

Deployment

Web appAPI

Integrations

Playwright Puppeteer Selenium Browser Use Stagehand Docker OpenAI Anthropic Gemini LangChain Claude Code Cursor Kernel.sh Steel Massive OpenAI CUA OpenClaw Cloudflare

Supported languages

English
Company

Behind Notte

Company name
Notte Labs, Inc.
Founded
25/09/2024
Country of origin
🇺🇸 United States
UBO
Andrea Pinto
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇨🇦 Canada
Legal contact
Support contact

Fundraising

Backed by Y Combinator, Summer 2025 batch, confirmed by the company's own footer link to its Y Combinator profile
No funding amount, round name or investor list is published on the site or on the Y Combinator page

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement Notte.

B BrowserbaseB BrowserlessB browser-useS SkyvernK Kernel
FAQ

Frequently asked questions

What exactly is Notte?
A browser infrastructure platform for AI agents. It provides cloud browser sessions, AI browser agents and serverless browser functions through a single API, together with credential vaults, browser profiles, proxies, CAPTCHA handling, session replays and observability for production web automation.
What does it cost to get started?
Nothing. The free plan costs 0 US dollars and grants 10 US dollars of credits once, with no expiry and no credit card required. The cheapest paid tier is Developer at 20 US dollars per month. Usage is then drawn from credits, at 0.05 US dollars per browser hour.
Do I need my own LLM API key?
No, it is optional. Notte is model-agnostic and works with OpenAI, Anthropic, Gemini and others. You can use Notte's keys at token price with no markup, or bring your own from the Startup plan upwards.
Can I keep using Playwright or Puppeteer?
Yes. Playwright, Puppeteer, Selenium, browser-use, Stagehand and anything that speaks the Chrome DevTools Protocol connect to Notte cloud sessions, in Python, TypeScript or Node.js. The site presents it as a one-line change with no rewrite.
Is Notte certified?
It is SOC 2 Type II certified and publishes a Trust Center listing 71 controls across business operations, cloud security, human resources and data, alongside the SOC 2 Type 2 report itself. HIPAA is offered on the Enterprise tier only.
Where is my data processed?
In the United States. The privacy policy states that Notte Labs is a United States company whose servers and central database operate there, and that international users consent to their information being transferred to and processed in the United States. No European or other regional hosting is offered.
Is Notte GDPR compliant?
The site never mentions the GDPR. Neither the privacy policy nor the terms of use contain the term, no Data Processing Agreement is offered and no Article 27 EU representative is named. The policy does address the EU AI Act and gates non-essential cookies for European Economic Area and United Kingdom visitors, and it grants CCPA and Nevada rights.
Is my data used to train AI models?
The site does not say. The privacy policy states that browser session data, meaning the pages visited, their content, URLs, screenshots and interaction logs, is used solely to run the requested automation and to provide session replay, then deleted according to the plan's retention period. It never takes a position on model training.
Is there a minimum age?
Yes, 13. Both the privacy policy and the terms of use state that Notte does not knowingly collect personal information from children under 13, and ask anyone younger not to use or register for the platform.
How do I get support?
By email at support@notte.cc, through a public Slack channel linked from the documentation, and via a status page for incidents. Legal matters go to legal@notte.cc, and a general contact address, hello@notte.cc, sits in the site footer.
Conclusion

Should you pick Notte?

Notte is a real product, well past the stage of a promising landing page. The documentation is deep, the SDK is public on GitHub, the status page and Trust Center work, and the SOC 2 Type II certification is backed by an actual report rather than a badge. For engineering teams that need software to drive a real browser at scale, the offer is coherent: sessions, agents and serverless functions behind one API, at 0.05 US dollars per browser hour, with a free tier generous enough to build something on.

Its sharpest idea is authentication. Substituting credentials at the browser-action layer keeps secrets out of the model's context window, and provisioning agents with real inboxes and phone numbers solves the sign-up and verification flows most competitors leave as an exercise for the customer. Existing Playwright or Puppeteer code connects over CDP without a rewrite, which lowers the cost of trying it.

The reservations concern governance rather than engineering. Notte Labs publishes no postal address, and its privacy policy invites you to write to a company address it never prints. More consequentially, the GDPR is absent from every legal document: no Data Processing Agreement, no Article 27 representative, no subprocessor list, and processing confined to the United States. The policy discusses the EU AI Act at length but never the regulation most European buyers will ask about, and it takes no position on whether customer data trains models. The legal pages themselves sit behind a bot challenge that keeps ordinary readers out.

The result is a tool that is easy to recommend to a United States engineering team and hard to clear through European procurement. Notte is young, and its compliance layer has visibly not caught up with its product.