Verifiable provenance
Every record carries its source URL, fetch timestamp and SHA-256. Every release ships a Merkle manifest signed via sigstore — check any record independently.
source_url · fetched_at · sha256
$ ragdata verify us-federal-procurement@2026.08.10 ✓ root matches
RAG-ready data.
Provenance you can verify.
Cryptographically verifiable, licensed-for-AI datasets from primary sources. Every record hashed, every release signed, every diff explicit — so your retrieval layer stands on data you can defend.
free sample datasets · no card · full provenance chain
merkle_root
9f2c41ab…c135e41a✓ signed release"why_ragdata":
Data you can put in front of your legal team
Licensed for RAG
An explicit RAG/embedding-use license instead of a gray area, with indemnification on Pro and Enterprise. Your legal review is one page, not forty.
embedding_use: permitted
Versioned releases with diffs
Chunk IDs are stable content addresses across releases. Pull the diff, re-embed only what changed, and keep your vector store bill flat.
+1,204 added · ~310 updated · −88 removed
Five niches live
US federal procurement, regulatory, courts, SEC filings, e-commerce. Collected from primary sources — never resold vendor feeds.
procurement ✓ regulatory ✓ courts ✓ sec ✓ ecommerce ✓
"how_it_works":
Pipeline, not promises
01 collect
We collect from primary sources
Official APIs and public portals — SAM.gov, FPDS, Federal Register, EDGAR. Raw responses are snapshotted immutably and hashed at fetch time.
02 normalize
We normalize and chunk with stable IDs
Cleaned, structured, chunked for retrieval. Chunk IDs are content addresses, so an unchanged chunk keeps its ID across releases.
03 release
You pull versioned releases via API
JSONL plus a signed manifest per release, with diffs between versions. Verify the Merkle root before a single byte enters your index.
"catalog":
Five niches, one contract
US federal procurement
RFPs, awards and attachments from SAM.gov and FPDS — normalized NAICS/PSC codes, linked award history.
sam.gov · fpdsdaily
Regulatory & compliance
Federal Register and EUR-Lex with versioned diffs — track what changed, not just what exists.
federalregister.gov · eur-lexdaily
Court records
State trial court dockets normalized across fragmented portals. Personal data stripped by design.
state court portalsweekly
SEC filings
EDGAR filings and earnings materials, chunked for retrieval with stable section anchors.
sec.gov / edgardaily
E-commerce products
Structured product facts for agentic commerce — attributes and availability, never copied prose.
public product feedsintraday
Your niche
Missing a corpus? We build custom niches and connectors on the Enterprise plan.
contact us →
"pricing":
Self-serve, per dataset
full pricing →demo
$0
no card required
- Sample datasets in every niche
- Full provenance chain on every record
- Signed Merkle manifest per release
- JSONL downloads
starter
$99/mo
per dataset / month
- Weekly release updates
- JSONL + signed manifest
- Per-record provenance fields
- Email support
most popular
pro
$299/mo
per dataset / month
- Daily release updates
- Versioned diffs, stable chunk IDs
- RAG/embedding license with indemnification
- Priority support
enterprise
Custom
custom agreement
- All datasets, all niches
- SLA on freshness and availability
- Custom niches and connectors
- Dedicated support
"faq":
Questions buyers actually ask
Can I legally embed and retrieve over this data?
Yes — that is the point. Every dataset ships under an explicit license that permits RAG, embedding, and retrieval use. Pro and Enterprise licenses include indemnification. No scraping gray zones: we collect from primary sources, honor robots.txt and opt-outs, and carry no personal data by design.
How do I verify provenance?
Every record carries its source URL, fetch timestamp, and SHA-256 of the raw response. Every release ships a manifest with a Merkle root over all record hashes, signed via sigstore. Recompute any record hash, walk it up the tree, compare against the signed root — independently, with the open-source verify CLI or your own code.
How often are datasets updated?
Cadence is per dataset and per plan: Starter ships weekly releases, Pro ships daily. Each release is versioned with a public changelog of added, updated, and removed records, so you always know what moved.
What formats do you ship?
JSONL for records and retrieval-ready chunks, plus a manifest.json per release with file hashes, counts, quality metrics, and the signed Merkle root. A Croissant (MLCommons) metadata file describes each release for dataset tooling.
Can I try before paying?
Yes. Demo access is free and needs no card — sign in with your email and download sample datasets in every niche, with the full provenance chain intact. Verify first, subscribe after.
Do you build custom niches?
On the Enterprise plan we build custom niches and connectors against your source list, with an SLA on freshness. Tell us what your retrieval layer is missing via the contact form.