Facts you can actually trust.
Every claim cross-checked against independent web sources, cited, dated, and confidence-scored before it goes live. Ask anything.
Featured facts
The Antikythera Mechanism used precisely cut, independently rotating gears in a system known as epicyclic gearing, a technology not otherwise known to exist in the West for another 1,500 years.
Archaeologists in Zambia found the world's oldest known wooden structure: two interlocking logs connected by a notch, dating to about 476,000 years ago.
Scientists revived a roundworm that had been frozen in permafrost for 46,000 years.
The marble tablet found in a New Orleans backyard was a gravestone for a Roman sailor.
A seaside town dealt with a giant whale carcass by using dynamite, which caused blubber dents on nearby cars.
The immortal jellyfish, Turritopsis dohrnii, can revert from adulthood back to its polyp stage when threatened by danger or starvation.
Abo Elementary School in Artesia, New Mexico was the first public school in the United States constructed entirely underground.
Venus rotates so slowly that a day on Venus (243 Earth days) is longer than a year on Venus (225 Earth days).
The guillotine was still in official use in France when Star Wars was released in 1977; the last execution by guillotine occurred in 1977, abolished in 1981.
Browse by topic
How DeepInquiry differs
Most AI-trust tools evaluate your data or your model's output and hand back a probability. That's useful. It's also not the same thing as knowing what's true. DeepInquiry sits one layer deeper: a verified fact corpus that AI systems can actually cite.
AI evaluation tools
Score model output
Statistical confidence on whether an AI answer seems correct. Runs on your data. Delivered inside a dashboard.
DeepInquiry
A verified fact corpus, served by API
Every claim cross-checked against independent sources, cited, dated, and tier-scored before it goes live. You get facts, not probabilities.
- We own the corpus.1,000+ facts and growing, each with independence-scored source clusters, citogenesis detection, and gold/silver/bronze verification tiers — not a service that runs on top of your data.
- Never a single source.Two or more structurally independent sources per fact. Wikipedia mirrors and content aggregators are excluded from independence scoring. If it can't be corroborated, it doesn't ship.
- Auditable, not opaque.Every fact response returns its full source cluster, independence score, cluster count, and verification tier. Governance teams get the provenance trail they actually need.
- Built for RAG, agents, and grounded UIs.A natural upstream source of truth for any grounded AI workflow: verified facts in, higher-quality outputs out.
Others score the AI. DeepInquiry is the ground truth AI can cite.
Cryptographically verifiable
Every fact returned by the DeepInquiry API carries a CiteSig™ attestation — an Ed25519 signature over the claim, its sources, and the time it was issued. Verifiers do not need to trust DeepInquiry: they check the signature against our published signer identity and get a mathematical yes/no.
- Signer identity is public.Fetch /api/v1/citesig/signer to get the DeepInquiry DID. Cache it, hard-code it, publish it. No auth required.
- The signature travels with the claim.Every fact response includes a
citesigobject withclaim,signer,sources,issued_at, andsig. Store it, forward it, verify it later. - Tampering is detected.Change any character of the claim or any source URL and the signature fails to verify. Even we can’t rewrite a signed fact after the fact — that’s the point.
- An open protocol, not a lock-in.CiteSig is an open standard with reference implementations in JavaScript and Python. Anyone can build a verifier; anyone can become a signer.
Signed, verifiable facts in your product
One API call, one confidence score, one set of citations, one cryptographic signature. Free tier includes 500 requests/month.
curl https://api.deepinquiry.ai/api/v1/facts/random \
-H "x-api-key: $DI_KEY"
# every fact response now includes a citesig
# signature block — verify with @citesig/core