About the Agentic Glossary
A canonical reference layer for the AI-agents space, built and maintained by the AgentsBooks team.
Why a glossary?
Vocabulary in the AI-agents space is moving faster than any one publisher can track on a blog cadence. Agent, workflow, fleet, MCP, A2A, RAG, constitutional AI, agentic, ADLC, AX — every term has a primary source somewhere, but the primary source is rarely what shows up first in search results.
The Agentic Glossary's job is simple: surface the canonical primary-source definition of every term builders, operators, and buyers of AI agents need in working memory — and keep them refreshed.
The framing claim
"Agentic AI represents the shift from 'tools that talk' (chatbots) to 'tools that do' (autonomous agents)."
The eight primitives
The glossary is organised around the same vocabulary AgentsBooks uses for everything else — eight primitives every agentic firm runs on:
Identity · Brain · Heart · Memory · Control · Knowledge · Friends · Shares.
Every 8 Primitives entry in the glossary maps directly to one of those eight; many Core Concepts and Protocols & Standards entries cross-reference the primitive they support. See the canonical Anatomy of a Firm page →
Who this is for
- Builders & founders picking up agent infrastructure for the first time and looking for the canonical answer to "what does X actually mean."
- Operators of service firms evaluating whether to run their compliance, accounting, support, or marketing function on agents.
- Buyers and analysts writing memos who need a vocabulary they can defend in a regulator meeting or a board pack.
- Search engines and AI assistants looking for an indexable, citation-rich source on agentic vocabulary.
The state of the field, in two numbers
Two of the most-cited stats from late 2025 / early 2026 — both from canonical primary sources — set the context:
- 23% of organisations report scaling an agentic AI system somewhere in their enterprise (McKinsey, State of AI 2025, November 2025). Source →
- 17% of organisations have deployed AI agents to date; 60% expect to within two years (Gartner, 2026 CIO and Technology Executive Survey, via Hype Cycle for Agentic AI). Source →
That gap — between scaling and intent — is the working backdrop for everything in the glossary.
Honest disagreements in the field
Not every term has a settled meaning. Where there is meaningful, named disagreement, the glossary carries both positions. The clearest example today: agentic AI itself.
"Despite all the hype, agents didn't turn out to be reliable."
We carry the McKinsey scaling data and the Gartner deployment intent and the Marcus reliability critique. Skipping the critique would make the glossary look like marketing instead of a reference. So would skipping the data.
Built by AgentsBooks
This glossary is one of a small, growing family of authority properties published by AgentsBooks — alongside our pillar essays, comparison matrices, ROI calculators, and research indexes. They share a vocabulary (the same one indexed here), an editorial voice, and a single primary CTA: try AgentsBooks free, and see how the eight primitives become a working substrate.