PDFs, reports, and filings → structured summary at your target token budget. NI 43-101 reports, SEDAR filings, press releases — compressed by GX10 and served as clean markdown + extracted fields.
Your agent needs to understand a 50-page NI 43-101 technical report or a 200-page SEDAR filing. DIGESTS compresses it to your target token budget — 200, 500, 2000, or full. Free tier: URL input only, 10-page limit. Paid: raw text upload, 100 pages, JSON field extraction.
Upload a URL or raw text. We extract the content (pdfplumber / pymupdf), then GX10 compresses it to your token budget. Structured markdown output.
Target 200, 500, 2000, or full. Agent picks the budget that fits its context window. Paid tier unlocks all 4 budgets.
Paid tier extracts structured fields: property name, commodity, measured + indicated ounces, inferred ounces. Output as JSON alongside the markdown summary.
Local GX10 (Blackwell) handles abstractive compression. 50x faster than cloud APIs. Degraded mode available with basic extraction if GX10 is down.
Compress a document by URL or raw text. Free: URL only, 10 pages, default budget. Paid: URL + raw text, 100 pages, all budgets, JSON fields.
curl -X POST "https://digests.4bots.ai/v1/digests/compress" \
-H "Content-Type: application/json" \
-d '{"url": "https://sedarplus.com/.../43-101-report.pdf", "budget_tokens": 2000, "style": "technical"}'
Example response:
{
"url": "https://sedarplus.com/.../43-101-report.pdf",
"title": "NI 43-101 Technical Report — Elk Gold Project",
"compressed_markdown": "## Property Description\n...\n\n## Mineral Resources\n...",
"token_count": 1847,
"original_estimated_tokens": 45000,
"compression_ratio": "96%",
"fields": {
"property": "Elk Gold Project",
"commodity": ["gold"],
"location": "BC, Canada",
"measured_and_indicated_ounces": 125000,
"inferred_ounces": 45000
}
}
Health check. Always free, always returns 200.
curl "https://digests.4bots.ai/health"
Free tier covers 5 documents/day (10 pages max). Paid tier unlocks 100-page documents, all token budgets, and structured field extraction.
No signup. No API key. Just send a request.
How x402 works