Reader

AI web infrastructure platform

Give your AI agent
the web context it needs

Turn URLs into clean Markdown, structured JSON, crawl results, and browser sessions so agents can read, research, extract, and act on live web content.

1,000 free credits every month. No card required.

Making the web readable for humans

curl -X POST https://api.reader.dev/v1/read \
  -H "x-api-key: $READER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com"}'
Response
# Example Domain

This domain is for use in illustrative
examples in documents.

You may use this domain in literature
without prior coordination or asking
for permission.

[More information...](https://www.iana.org/domains/example)

One API for the web context agents keep needing

Reader helps your agent work with the web in four core ways.

Scrape a page

Send Reader a URL and get back clean page content. Markdown for model context, HTML for inspection, metadata for storage, screenshots for visual checks, and raw page content when your system needs lower level access.

Crawl a website

Start from one URL and collect multiple pages from a website. Control depth, limits, and URL patterns so your crawler collects the pages that matter without turning every task into an open ended crawl.

Extract structured data

When your workflow needs fields instead of full text, Reader can extract structured JSON from web content. Define the data shape you want with a schema or instruction, and get usable fields back with the page context.

Open a browser session

Some pages need real interaction. Reader Browser lets your workflow launch a cloud browser session and connect with automation tools such as Playwright or Puppeteer for clicks, navigation, forms, screenshots, PDFs, and rendered state.

Built for real agent workflows

Agent research

Let an agent read public websites, crawl related pages, extract structured fields, and summarize findings with source context.

RAG ingestion

Convert documentation, blogs, help centers, and public websites into clean Markdown for chunking, embedding, retrieval, and answer generation.

Product data workflows

Collect product names, prices, availability, variants, reviews, and metadata from public product pages.

Price monitoring

Retrieve product pages, extract price and availability fields, store snapshots, and compare changes in your own system.

Competitive intelligence

Crawl public competitor pages and feed clean content into your own change detection, summarization, and alerting workflows.

Lead research

Collect public company descriptions, product pages, location information, contact page links, hiring signals, and website content for research workflows.

Solutions for every team

Monitor product prices, track inventory, and extract product data across thousands of stores. Reader handles JavaScript rendering and anti-bot protection on e-commerce platforms automatically. One API call returns clean, structured data from any product page.

Learn more →

Why teams choose Reader

It is built around agent context

Reader is designed for workflows where web content becomes model context, structured data, crawl results, or browser controlled state.

It keeps the product model simple

Scrape pages, crawl websites, extract structured data, and open browser sessions without stitching together separate services.

It returns useful outputs

Get Markdown for models, JSON for applications, metadata for storage, HTML for inspection, and browser sessions for interactive workflows.

It works across use cases

Use the same platform for agent research, RAG ingestion, product data, price monitoring, competitive intelligence, lead research, and website analysis.

It gives developers control

Use the API, SDKs, CLI, MCP server, Playwright, Puppeteer, webhooks, and your own storage or agent framework.

No infrastructure to manage

No browsers to run, no proxies to rotate, no servers to scale. Reader handles rendering, anti-bot bypass, and content extraction so your team builds the product, not the plumbing.

Works with your stack

PythonJavaScriptPlaywrightPuppeteerLangChainLlamaIndexCrewAIClaude CodeCursorVS CodeZapierMake

How It Works

Three steps. That's it.

01

Give Reader a URL, or connect to a cloud browser

A single page, an entire site, or a live browser session. Reader handles all three.

02

Reader handles the hard part

Rendering, anti-bot bypass, content extraction, session management.

03

Get clean, structured data

Through the platform or the API. Ready for your agents, your workflows, or your next report.

Frequently asked questions

Start with your first URL

Give your agent the web context it needs. Scrape a page, crawl a site, extract structured data, or open a browser session from one API.

Start free