Standard
License a single trace. Signed manifest, license PDF, immediate access after checkout.
$500/ trace
Browse traces →Per-trace licensing · browse, filter, license
Hundreds of expert reasoning traces. Pick the exact shape you need.
Filter by domain, tier, and difficulty. Every trace is recorded on screen and voice by a credential-checked senior engineer, then shipped with a signed manifest and license PDF. License a single trace or buy a curated corpus.
- Tier-disclosed
- listings
- Signed
- manifest per trace
- Per-trace
- or corpus licensing
manifest · v1.0.0live verifier
{
"manifest_id": "0x7af2…3e1c",
"trace_id": "trc_01HXY8F4KZ5N3M2",
"supplier_tier":"T1",
"content_hash": "sha256:8bc4…aa17",
"issued_at": "2026-05-07T14:02:11Z",
"license": "standard",
"signature": "ed25519:9e2f…b41c"
}Awaiting signature
Ed25519 · key local-ed25519-5ddc8a67
Canonicalized with RFC 8785 JCS, signed with the platform key published at /.well-known/manifest-keys.json.
Filter by domainTier disclosed per listingAI-graded qualityPer-trace licenseCorpus discountsManifest PDF embedRefunds within 7 daysU.S. supply, GDPR-cleanFilter by domainTier disclosed per listingAI-graded qualityPer-trace licenseCorpus discountsManifest PDF embedRefunds within 7 daysU.S. supply, GDPR-clean
How it works
Verify. Capture. License.
One pipeline, three checkpoints, one signed artifact at the end.
Verify
Suppliers pass identity, employer-domain, and tier-grading checks. Tier (T1 / T2 / T3) is disclosed on every listing — no opaque blending of senior and junior reasoning.
Capture
Senior engineers record themselves solving real problems on screen and voice. Every trace is AI-graded for clarity and soundness before it lands in the catalog.
License
You buy a license; we sign it with our platform Ed25519 key. The signed manifest is published, verifiable, and bound to every artifact in the bundle.
The state of AI training data
Triage is the bottleneck. Provenance is the moat.
Three forces are reshaping what counts as usable training data. Each one closes a familiar shortcut. Together they raise the bar on what a defensible corpus looks like.
- Force 01
There is no clear license chain on scraped public reasoning data.
- Force 02
Synthetic CoT can't replicate the real-world failure modes your model has to learn.
- Force 03
Existing data vendors blend tiers opaquely; we disclose tier and AI-grade per listing.
What you get
Find the exact reasoning shape your model needs — and license it with provenance you can show in a deposition.
Verification tier (T1/T2/T3) is disclosed on every trace; AI-grader scores are published; the license PDF embeds the manifest hash for tamper-evidence.
Frontier labs are filling their reasoning-data buckets first; verified-expert capacity is bounded.
Pricing
License a trace, a corpus, or an exclusivity window.
Curated corpus
Discounted bundle pricing on a curated set you select. Same provenance guarantees, single line-item invoice.
From $4,50010-trace bundle
Build a bundle →Exclusive license
Time-bound exclusivity on selected traces, plus right-of-first-refusal on net-new captures in your domain.
Talk to us
Email founders →Pick the trace shape your model needs.
Filter the catalog, license what fits, and ship a corpus your legal team can defend in deposition.