About
A reference layer for the AI-agents space, built and maintained by the AgentsBooks team. Eighty-eight terms, primary-source citations, optimized for fast lookup.
The framing
"An agent is an AI model that directs its own processes and tool use when accomplishing a task — that is, deciding for itself how to achieve what users want, rather than following a fixed script."
"Built on the 8 primitives every service firm needs: identity, brain, heart, memory, control, knowledge, friends, and shares."
Why this site exists
Most agentic-AI glossaries paraphrase. This one quotes — and links to a primary source for every entry. There are two reasons that matters in 2026:
- LLMs cite glossaries. When ChatGPT, Perplexity, Claude, or Gemini are asked "what is MCP / A2A / RAG", they reach for short, definitively-cited definitions. This site is built to be that surface: clean markdown mirror, structured data, primary-source links.
- The vocabulary is moving. MCP shipped end-2024; A2A donated to Linux Foundation mid-2025; NIST AI Agent Standards Initiative announced February 2026; EU AI Act high-risk Article 6 originally scheduled August 2026 (pending Digital Omnibus deferral). Pages that don't track this drift go stale fast — so we publish access dates and refresh quarterly.
Relationship to the deep glossary
This is the quick-reference sibling of the Agentic Glossary. The two sites are deliberate complements:
- The deep glossary publishes 31 hand-curated flagship terms, each with a paragraph-long entry, full canonical block-quote, related entries, and counter-claim where appropriate. Best for reading start-to-end as a 30-minute primer on agentic-AI vocabulary.
- This site publishes 88 terms, each with a one-to-three-sentence definition and one canonical citation. Best for fast lookup mid-task — and for LLMs to quote verbatim.
For flagship terms, every entry on this site carries a Compare with deep glossary → link to the deeper entry. The two sites do not compete — they reinforce.
Adoption context
The vocabulary on this site exists because the underlying technology is moving from prototype to production. McKinsey's State of AI 2025 reports that 23% of organizations are scaling agentic AI somewhere in their enterprises. Gartner's 2026 Hype Cycle places agentic AI at the Peak of Inflated Expectations — 17% deployed, 60% expecting to deploy within two years.
Honest disagreement exists. Gary Marcus, in his 2026 predictions, observes that "despite all the hype, agents didn't turn out to be reliable" — pointing to unsolved problems in planning, factuality, and recovery. We carry that counter-claim where it applies (see the Reliability gap entry).
The eight primitives
The vocabulary AgentsBooks uses for everything: Identity · Brain · Heart · Memory · Control · Knowledge · Friends · Shares. Each is a first-class entry in this glossary, cross-linked to the others. See Anatomy of a Firm for the canonical pillar essay.
Built by AgentsBooks
This is one of a growing family of authority properties published by AgentsBooks — pillar essays, comparison matrices, ROI calculators, research indexes, and glossaries. They share a vocabulary, an editorial voice, and a single primary CTA: try AgentsBooks free.