bluefield

What takes months to build. Ready in minutes.

A production web data API at a fraction of the cost and time of building your own fetcher, renderer, and cleanup stack. Every live fetch returns a signed, offline-verifiable capture. US hosted with GDPR aware defaults.

  • Signed, verifiable captures
  • US hosted · GDPR aware

1,000 free credits per month · Upgrade to Pro at $19 when you are ready

See pricing →
Try it

See it yourself

Paste any URL and get clean data back — no signup, no API key.

Examples

Pick an example or paste a URL

The example buttons show curated output. Paste any other URL to run a live scrape.

Messy web. Clean data.

The same URL. Two very different results.

BeforeRaw HTML
<div class="article-wrapper noscript-hidden"
  data-reactroot="">
  <script>
    window.__INITIAL_DATA__={"ads":true}
  </script>
  <span class="css-1a2b3c hydrate">
    <!-- react hydration marker -->
  </span>
  <div class="MuiBox-root css-xyz">
    <div class="ad-slot" data-slot="top">
    </div>
    <p class="MuiTypography-body1
        makeStyles-text-42">
      The article content
      is buried here...
    </p>
  </div>
</div>
AfterClean markdown
# Building a RAG Pipeline in 2025
 
Published June 2025. From example.com.
 
A retrieval augmented generation system
connects your language model to live data.
Here is how it works.
 
## Three key components
 
- A retrieval source with fresh content
- A language model to reason over it
- A pipeline to connect them together

Everything your AI needs from the web

Four primitives. One API key. Any data source.

Ask any question and get full page content back, not just links or snippets. Ready for your model or agent to read.

Search results

Give one URL and receive clean markdown, JSON, or a screenshot. No HTML parsing or rendering required on your side.

Clean markdown

Discover all URLs on a site then crawl and extract each page at scale. One API call starts the whole job.

All pages
# Scrape a page to clean markdown
curl -X POST https://api.bluefields-data.com/scrape \
  -H "Authorization: Bearer $BF_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url": "https://news.ycombinator.com", "formats": ["markdown"]}'
Build vs buy

Less glue code. More shipping.

Building it yourself means a browser fleet, proxy rotation, extraction rules, and a chunking step. Send one request and get clean data back.

See the comparison →
Agent ready

Give your agents access to the live web

Register Bluefield as an MCP server in one command. Your agent gets search, scrape, map, and crawl as native tools it can call the same way it calls any other function.

One call to start

curl -X POST https://api.bluefields-data.com/scrape \
  -H "Authorization: Bearer $BF_KEY" \
  -d '{"url": "https://example.com"}'
✓  clean markdown returned  ·  ready for your model

MCP server config

{
  "mcpServers": {
    "bluefields": {
      "command": "npx",
      "args": ["-y", "@bluefields/mcp-server"],
      "env": { "BF_KEY": "your-key-here" }
    }
  }
}
✓  bluefields MCP  ·  5 tools registered in your IDE

Or use the SDK

Read the SDK docs ↗
import { BlueFieldsClient } from '@bluefields/sdk';

const client = new BlueFieldsClient({ apiKey: process.env.BF_KEY });
const r = await client.scrape({ url: 'https://example.com' });
console.log(r.markdown);
✓  r.markdown  ·  clean content, no HTML parsing
Live data

Watch a page. Get told when it changes.

Register a watch on any URL. When the cleaned content changes we send you the diff through a webhook. No polling and no scraping on your side.

Register a watch

One POST. We track the page from then on.

curl -X POST https://api.bluefields-data.com/subscriptions \
  -H "Authorization: Bearer $BF_KEY" \
  -d '{"url":"https://example.com/pricing","webhook_url":"https://you.dev/hook"}'

Change event

Delivered to your webhook the moment content changes.

POST https://you.dev/hookWebhook payload
{
  "event": "changed",
  "url": "https://example.com/pricing",
  "diff": {
    "added": ["Pro plan $29/mo"],
    "removed": ["Pro plan $19/mo"]
  }
}
Verifiable by default

Signed results you can prove weren't tampered with.

Every live fetch returns a signed, offline-verifiable capture. Cached responses and PDFs are unsigned in v1.

Signed capture

Every live /scrape returns a signature and signed_at.

{
  "extraction_id": "ext_8f3a91c4",
  "url_canonical": "https://example.com/pricing",
  "content_hash": "sha256:9f2b7c4e…a1d0",
  "attestation": {
    "algorithm": "EdDSA",
    "key_id": "bf-key-2026-01",
    "signature": "MEYCIQ…base64url…",
    "signed_at": "2026-05-31T08:14:22Z"
  }
}

Verify offline

One command checks the signature with no call back to us.

$ npx bluefields verify response.json
✓ signature valid · content unmodified
Compliance

Built for teams that need to defend their data choices

We state what is live today and what is not. No certification badges we have not earned and no residency claims we cannot support yet.

Hosted in the United States today

The managed API runs on Fly.io in us east (iad). We do not claim EU data residency on the default hosted tier.

GDPR aware by design

We respect robots.txt by default. You choose which URLs to fetch and remain responsible for lawful basis and purpose. A data processing agreement is available on request.

Signed, verifiable captures

Every live fetch returns a cryptographic manifest you can verify offline — one-shot /scrape and watch mode alike. Cached responses and PDFs are unsigned in v1.

Honest about certifications

We do not hold SOC 2 or ISO 27001 today. Enterprise and self hosted options exist for teams with specific compliance needs.

Need EU region hosting, a DPA, or a custom deployment? Talk to us about enterprise options. This page is not legal advice.

01

Send a URL or a question

Paste any web address or type a question in plain English.

02

We render and clean

JavaScript rendering, bot detection, and noise removal handled automatically.

03

Receive clean structured data

Get back markdown, JSON, or a screenshot ready to use in your pipeline.

Zero configuration

We handle the hard parts

Everything that takes weeks to build yourself, ready in one API call.

Reading report.pdf

Web pages and PDFs, parsed properly.

Link any page or PDF. Get clean structured text back.

Requesting…

Waits for JavaScript

We detect when content has loaded, not just when HTML arrives.

Instant repeat requests

Repeated URLs return cached clean data in milliseconds.

Navigate
Click
Type
Wait
Scroll
Screenshot

Interact before you extract

Click, scroll, type, or wait. Automate the page before scraping.

Respectful by default

We honor robots.txt and rate-limit our own requests. Public surface only — we never log in to sites or bypass paywalls.

Integrations

Works with your stack

Drop Bluefield into any AI framework or pipeline. The same API returns clean markdown, structured JSON, or a screenshot depending on what you ask for.

See all integrations

Compatible frameworks

LangChain
LlamaIndex
OpenAI
Anthropic
Vercel AI
CrewAI
AutoGen
MCP
Proof, not promises

Numbers you can check yourself

We publish methodology and reproduction steps. Headline benchmark figures are held until the latest re-run is reviewed. The product itself you can verify, price out, and run on your own.

Benchmarks

Benchmark figures are held until the latest competitive re-run is reviewed. The /benchmarks page describes our stratified corpus, method, and caveats.

You do not have to take our word for it.

Every live fetch returns a signed, offline-verifiable capture. Cached responses and PDFs are unsigned in v1. Verify any capture offline with the published spec and the npx bluefields verify CLI. Webhook deliveries are signed the same way.

See how verification works →

Built in the open

The fetcher, extractor, differ, and benchmark are all readable.

Read the source →

Published pricing

Every operation has a price you can see before you call it.

See pricing →

Self-hostable

Run the same API on your own infrastructure with Docker Compose.

Read about self-host →
Datasets

Request a curated dataset

One time and subscription dataset delivery is coming soon. Tell us what you need and we will reach out when it is available.

Request a dataset
Use cases

Turn web data into AI powered products

Web tools your agents can call natively

Register Bluefield as an MCP server. Your agent gets search, scrape, map, crawl, and extract as native tools it can call the same way it calls any other function.

Common questions

Frequently asked questions

Bluefield is a web data API built for AI. Send a URL or a search query and receive clean structured data. Markdown, JSON, or screenshots, ready to use in your models, agents, or pipelines.

Yes. A Docker image is available so you can deploy Bluefield inside your own cloud account. Enterprise plans include deployment support and an uptime commitment.

Developers building AI applications that need reliable web data. Research agents, RAG pipelines, chatbots, lead enrichment tools, and anything that needs to read the live web.

The fastest way to make the web AI ready

Free tier available. No credit card required.

npx bluefields init
Deploy it in your own cloudAll systems operational