Nobody publishes machine-readable AI governance disclosure yet. Including us.
The first real crawl of the AI Procurement Pulse is in. We probed 37 vendor domains across
9 verticals for all eleven Kinetic Gain Protocol Suite
documents at their /.well-known/ paths. The result is unambiguous — and honest.
This is the starting line, not a verdict
The Suite shipped in 2025. Machine-readable AI vendor disclosure is a brand-new practice, and a brand-new practice begins at zero. Issue #1 exists to calibrate the instrument and mark that zero — so every future issue can measure the climb against a real, reproducible baseline rather than a vibe.
We held ourselves to the same ruler. Kinetic Gain's own properties — kineticgain.com,
suite.kineticgain.com, clinical.kineticgain.com, decisions.kineticgain.com —
are in the universe, and they score zero too. You can't run a disclosure index where the author is exempt.
Dogfooding our own specs is the first item on the Issue #2 docket.
200 OK serving unrelated JSON at a well-known path is not counted
as a published spec. Every domain genuinely publishes nothing at these paths today.
The universe — by vertical
37 domains, hand-selected to span the AI buying market. Publication rate is the share of domains in each vertical publishing any Suite document.
| Vertical | Domains | Publishing | Rate |
|---|---|---|---|
| AI Platform | 9 | 0 | 0.0% |
| EdTech | 7 | 0 | 0.0% |
| HealthTech | 5 | 0 | 0.0% |
| Enterprise SaaS | 4 | 0 | 0.0% |
| Reference (Kinetic Gain) | 4 | 0 | 0.0% |
| FinTech | 3 | 0 | 0.0% |
| Data | 2 | 0 | 0.0% |
| Identity | 2 | 0 | 0.0% |
| Observability | 1 | 0 | 0.0% |
Domains include OpenAI, Anthropic, Google, Microsoft, Khan Academy, Duolingo, Tempus, PathAI, Stripe, Salesforce, ServiceNow, Snowflake, Okta, CyberArk, and more. Full universe + raw results: issue-1/data.json · procurement-pulse-engine.
Per-spec adoption
All eleven specs were probed at every domain. Every cell is zero — this table is the scoreboard future issues fill in.
| Spec | Well-known path | Publishers |
|---|---|---|
| AEO Protocol | /.well-known/aeo.json | 0 / 37 |
| Agent Cards | /.well-known/agents/ | 0 / 37 |
| Prompt Provenance | /.well-known/prompts/ | 0 / 37 |
| AI Evidence Format | /.well-known/evidence/ | 0 / 37 |
| MCP Tool Cards | /.well-known/tool-cards/ | 0 / 37 |
| AI Tutor Cards | /.well-known/tutor-cards/ | 0 / 37 |
| Student AI Disclosure | /.well-known/student-ai/ | 0 / 37 |
| Classroom AI AUP | /.well-known/aup.json | 0 / 37 |
| Clinical AI Disclosure | /.well-known/clinical-ai/ | 0 / 37 |
| AI Incident Card | /.well-known/incidents/ | 0 / 37 |
| AI Procurement Decision Card | /.well-known/decisions/ | 0 / 37 |
What changes this number
Adoption climbs when buyers start asking for it. A district that publishes an AI Procurement Decision Card forces the vendors it reviews to publish the declarations that card cites. The fastest path off zero is procurement pressure, not vendor goodwill.
Roadmap. Issue #2 expands the universe toward the full ~1,200-domain target (Fortune 500 + top 100 K-12 EdTech + top 50 HealthTech AI), adds signature verification, and reports the first drift deltas. And it will show Kinetic Gain's own properties as the first non-zero row — we publish before we ask anyone else to.