Deep crawl the web with Claude Sonnet 4
Intelligence.
Send schema & sites → Get structured data. No
brittle selectors.
Response includes confidence scores and source URLs for every extracted value
POST https://api.eddie.surf/crawl
{
"urls": ["https://www.fijiwater.com", "https://www.evian.com", "https://www.volvic.com"],
"callback_url": "https://webhook.site/your-unique-id",
"max_depth": 2,
"max_pages": 10,
"context": {
"purpose": "Evaluating for supplying our sustainable cafe"
},
"rules": ["Don't crawl shop pages"],
"json": {
"company_name": {
"type": "string",
"description": "Company name",
"additional_guidelines": ["Don't include suffixes like Ltd, Inc, Pte, etc"]
},
"sustainability_score": {
"type": "score",
"description": "Environmental sustainability rating (1-10)",
"additional_guidelines": ["10=carbon neutral + 100% recyclable packaging, 8-9=100% recycled materials + energy efficiency, 6-7=recycled packaging initiatives, 4-5=basic environmental programs, 1-3=minimal sustainability efforts"]
},
"publicly_traded": {
"type": "boolean",
"description": "Whether company is publicly traded"
},
"employee_count": {
"type": "number",
"description": "Approximate number of employees"
}
}
}
200 OK
{
"status": "completed",
"pages_crawled": 8,
"results": {
"company_name": {
"value": "FIJI Water",
"confidence": 5,
"whereFound": "Found at [1]\nSources:\n[1] https://www.fijiwater.com/ - Homepage header and branding"
},
"sustainability_score": {
"value": 8,
"confidence": 4,
"whereFound": "Score based on '100% recycled plastic bottles', 'energy efficiency initiatives', and 'microturbine energy generation' from [1] and [2]\nSources:\n[1] https://www.fijiwater.com/sustainability - Environmental commitments\n[2] https://www.fijiwater.com/about-us - Sustainability programs"
},
"publicly_traded": {
"value": false,
"confidence": 3,
"whereFound": "No mention of public trading found across crawled pages"
},
"employee_count": {
"value": 500,
"confidence": 3,
"whereFound": "Estimated from company size mentions and facility descriptions"
}
}
}
Cheaper, and more accurate than alternatives
⚠️ Do not use HTML-to-markdown tools
They lose critical context that AI needs for accurate extraction.
Why? See FAQ below →Developers are using Eddie to power their next-gen applications
Give your agents real-time web data access. Eddie surfaces the information they need without getting stuck on changing page structures.
Automate prospect research by extracting company info, contact details, and tech stacks from websites at scale.
Monitor competitor pricing, product updates, and industry news by surfing multiple sources automatically.
Integrate with n8n, Zapier, and Make.com workflows. Automate data collection for CRM updates, lead scoring, and market monitoring.
We use proprietary optimization techniques:
• Smart utility cleaning: Our LLM trained on hundreds of frameworks strips utility classes (like "btn-primary-lg-mobile") while preserving semantic ones (like "news-article" or "product-price")
These optimizations, combined with infrastructure efficiencies, achieve the $0.04 per page price point.
Standard markdown converters lose critical context that our AI needs:
• Visual Relationships: We preserve layout context so data points like dates stay aligned with correct information such as articles in listings
• Semantic Structure: Keeps grouped data together (price + product + description) rather than flattening everything
• Multi-Page Synthesis: Combines information across pages to provide complete, verified answers
• Source Attribution: Every extracted value includes the exact page URL where it was found
Our AI doesn't just follow every link—it intelligently prioritizes based on your context:
• Context-Aware Scoring: Links are ranked by relevance to your extraction goals. For example, "Home > Marketing services > Meet the team > Joe Bloggs" provides much more meaningful context than just finding the "Joe Bloggs" page alone
• Adaptive Depth: Automatically stops when sufficient data is found or diminishing returns detected
• Content-Type Detection: Skips irrelevant pages or content based on your rules
Everything you need for production data extraction:
• Smart Crawling: AI-powered link discovery and prioritization
• Data Processing: Claude Sonnet 4 analysis with confidence scoring
• Multi-Page Synthesis: Combines data across pages for complete results
• Source Attribution: Detailed provenance for every extracted value
• Infrastructure: Proxies, rate limiting, retry logic, and callback delivery
No hidden fees, setup costs, or per-token charges. One simple price per page crawled.