Glossary
Words mean things — especially in a discipline that claims to be measurable. This is the canonical vocabulary we operate: every term below means exactly this, everywhere we use it — on this site, in the diagnostic, in the research.
- AI-native organization
- An organization where humans and AI agents work, learn, and evolve together under working institutions — reliably, predictably, measurably. Not "a company that uses AI tools": the difference is whether the organization's capabilities compound, and whether anyone can measure that they do.
- AI-nativity
- The degree to which an organization is AI-native. Not a label or an aspiration — a measurable position on the L0–L5 ladder, graded per capability area, with evidence behind every placement.
- The L0–L5 ladder
- The graded maturity scale of AI-nativity. A placement is made per capability area against the technology map — never as a single flattering number. The scale is strict by design: by it, we ourselves score L0.9 today (self-assessment). An easy scale would just tell everyone they're fine. (Not to be confused with the agent-level L0–L5 trust ladder in our accountability architecture — that scale grades individual agents; this one grades organizations.)
- Evidence class
- The grading attached to every technology claim: deployed & measured (running in production, with numbers) · working prototype · specified · frontier (open research). Claims and evidence travel together — on this site too.
- The technology map
- A graded tree of the 38 technologies that make an organization AI-native — 13 deployed & measured, 14 prototyped, 9 specified, 2 frontier — spanning the technical zone and the human zone (roles, learning, trust). The backbone the read is derived from.
- Coverage
- The set of functions an AI-native organization has to close — from agent accountability to human–agent trust. The coverage schema maps what 24 published standards (NIST, ISO 42001, OWASP and others) cover, what they partially cover, and the institutional layer none of them closes.
- Coverage schema
- The artifact that maps 24 published standards against the coverage functions — what each covers, what it partially covers, and the institutional layer none of them closes. Read alongside the technology map.
- Trust = Alignment × Reliability
- Trust in an agent is a product, not a sum. Alignment: is it pointed the right way? Reliability: does it behave as expected, every time? A perfectly aligned agent that fails randomly is untrustworthy — and a perfectly reliable one pointed the wrong way is worse. Zero in either term makes trust zero.
- Agent reliability
- An agent's behavior staying within expected bounds in production — measured, not assumed. Reliability failures are not only model failures: roles, learning, and trust break agents too (the essay).
- HALA — Human-AI Learning Architecture
- Our universal architecture for how knowledge is taught — to people and agents alike. The same framework runs in companies, universities, and training programs; Agents University is built on it. Full framework — white paper on request.
- The read (the AI-nativity diagnostic)
- A fixed-scope, evidence-grounded assessment: your L0–L5 placement per capability area, your named coverage gaps, and one defensible next move — not a 40-item roadmap, and not the front-end of a transformation contract. How it runs.
- The self-check
- A five-minute, self-serve preliminary placement on the L0–L5 ladder — no call, no commitment. The low-barrier door into the discipline. Currently in final validation (waitlist open).
- Agents University
- The teaching arm of the 8Hats ecosystem — where humans and agents learn the discipline the lab measures. Runs on HALA. agents.university
A term used somewhere on this site and missing here is a bug — tell us.