Overview
Anything API is a product by Notte Labs that converts any browser-based workflow into a production-ready API endpoint — automatically. You describe the task in plain English, and Anything API's AI agent reverse-engineers the workflow, builds a custom Notte Function, and exposes it as a callable API.
Most websites don't offer public APIs. Developers who need to extract data or automate repetitive actions have traditionally had to maintain brittle browser scripts. Anything API eliminates that maintenance burden: after an initial browser session figures out the workflow, every subsequent call skips the browser entirely and hits the target site directly — making repeated runs faster and cheaper.
It's aimed squarely at developers, AI engineers, and data teams who need reliable, scalable web automation without writing and maintaining bespoke scrapers. The free tier includes 100 browser hours with no credit card required, making it easy to validate use cases before committing.
Key Features
Natural Language → API Endpoint — Describe any web task (e.g., "Get price and reviews for any Amazon product") and Anything API builds a callable REST endpoint that returns structured JSON, eliminating the need to write automation scripts from scratch.
Browser-Free Subsequent Runs — The initial session does the heavy lifting of figuring out the workflow. Every subsequent call bypasses the browser and calls the target site directly, dramatically reducing latency and infrastructure cost.
Anti-Detection & CAPTCHA Handling — Built-in residential proxies and CAPTCHA solving (coverage and success rate vary by site and scenario) help agents operate across most websites. Effectiveness depends on site complexity and bot-detection strength. Users are responsible for ensuring automations comply with each target site's Terms of Service and applicable data regulations.
Serverless Deployment via Notte Functions — Generated APIs deploy as serverless functions colocated with the browser infrastructure. Schedule them on cron, invoke them via API, or trigger them from any AI workflow automation stack.
Secure Authentication Handling — Agent Vaults store credentials encrypted at rest and in transit — secrets are never sent to LLM providers. Agent Identities supply dedicated emails and phone numbers for OTP/2FA flows (Persona identities incur additional charges), and Session Profiles persist full browser state so agents skip re-authentication on every subsequent run.
Model-Agnostic LLM Usage — Works with OpenAI, Anthropic, and Gemini. You can bring your own API keys or use Notte's — billed at token price with no markup, making it cheaper than alternatives that charge per-step fees.
Pricing & Plans
Anything API runs on Notte's credit-based billing system. Plans primarily control rate limits (concurrency and browser hours); actual usage is billed on top.
| Plan | Base Price | Browser Hours | Concurrency | Stealth |
|---|---|---|---|---|
| Free | $0/mo | 100 hrs (forever) | 5 | — |
| Developer | $20/mo | 100 hrs/mo | 25 | Basic |
| Startup | $100/mo | 500 hrs/mo | 100 | Advanced |
| Enterprise | Custom | Usage-based | 250+ | Advanced+ |
Usage costs (all plans):
- Browser usage: $0.05/hour (charged per minute, rounded up)
- Proxy usage: $10/GB (premium residential proxies included)
- LLM/Agent steps: Pay token price — no per-step markup
- Persona identities: Additional charge applies for dedicated email and phone services
For context, a typical 1-minute agent run using Gemini 2.0 Flash with US proxies costs approximately $0.116 total (per Notte's official example). Compared to alternatives like HyperAgent ($0.02/step) or Browser Use (agent steps from $0.002/step), Notte's per-token billing can be more cost-effective for token-efficient tasks where proxy traffic is low — though proxy costs can dominate in high-volume scraping scenarios.
The Free plan requires no credit card and is a genuine permanent tier — not a trial.
Best For
- Developers who need to automate web workflows on sites with no public API
- AI engineers building agentic pipelines that require structured web data at runtime
- Data teams running recurring extraction jobs (pricing monitors, competitive intelligence, listing aggregators)
- Startups that want API-quality reliability from browser automation without maintaining scraper infrastructure
- Teams already using Playwright, Puppeteer, or Selenium who want a managed, scalable layer on top
FAQ
What is Anything API?
Anything API is a service by Notte Labs that turns any browser workflow into a callable API endpoint. You describe a task in plain text, and the system's AI agent builds and deploys a Notte Function that executes the workflow on demand, returning structured JSON.
How is Anything API different from a standard web scraper?
Traditional scrapers require you to write and maintain code against a specific site's HTML structure. Anything API uses an AI agent to figure out the workflow automatically, and on subsequent calls bypasses the browser entirely — making it faster and more resilient to minor page changes.
Is there a free plan, and what does it include?
Yes. The Free plan provides 100 browser hours as a permanent allocation (not a trial), up to 5 concurrent sessions, with no credit card required to sign up. Note that proxy traffic ($10/GB) and LLM token costs are billed separately even on the free tier — actual charges depend on your usage pattern.
How does Anything API handle websites that require login?
Through three mechanisms: Agent Vaults store credentials encrypted at rest and in transit without exposing them to LLMs; Agent Identities provide dedicated email addresses and real phone numbers for OTP/2FA verification (Persona identities are an additional billed service); and Session Profiles save full browser state so agents can skip the login flow on every subsequent run.
Which programming languages and frameworks are supported?
Anything API outputs Notte Functions, which are compatible with Python, TypeScript, and Node.js. The underlying browser infrastructure integrates with Playwright, Puppeteer, Selenium, browser-use, and Stagehand via the Chrome DevTools Protocol (CDP).
How does the LLM billing work?
Notte charges the actual token price of whichever model you use — with no markup. You can bring your own OpenAI, Anthropic, or Gemini API keys, or use Notte's credentials and be billed at cost. This contrasts with competitors that charge a flat per-step fee regardless of token consumption.
Is Anything API SOC 2 compliant?
Yes. Notte (the underlying platform) is SOC 2 Type II certified. Security details are available at their Trust Center.
Can I schedule Anything API functions to run automatically?
Yes. Generated Notte Functions can be deployed serverless and scheduled via cron, invoked via API on demand, or triggered from external workflow tools. The Notte Studio provides a visual IDE for editing, testing, and one-click deployment.
