Sunday, June 7, 2026

The AI Buzzword Glossary: From Agents to Agentic AI

The AI Buzzword Glossary: From Agents to Agentic AI

Every week brings a fresh wave of AI terminology. “Agentic AI,” “RAG,” “MCP,” “context engineering” — the words pile up faster than anyone can keep track of them, and half the time they’re used loosely or interchangeably.

This post is a no-jargon reference. Each term gets a plain-English definition (what it is) and a short note on what it’s for — so you can read a vendor blog or a conference talk without nodding along to words you don’t actually use yourself.

The terms are grouped by theme so related ideas sit together.


1. Core Agent Terminology

This is the cluster everyone is talking about — the shift from chatbots that answer to systems that act.

TermWhat it isWhat it’s for
AI AgentAn LLM-powered system that can take actions — call tools, hit APIs, run code — to accomplish a goal, not just produce textAutomating multi-step tasks like data lookups, ticket resolution, or bookings
Agentic AIThe broader paradigm where AI plans, decides, executes, and self-corrects with some degree of autonomyThe umbrella term for moving from “assistant that talks” to “system that does”
Autonomous AgentAn agent that runs with minimal or no human input, looping until its goal is metBackground jobs, monitoring, “set and forget” workflows
Multi-Agent System (MAS)Several specialized agents collaborating, each owning a sub-taskComplex workflows where one agent coordinates others (planner + researcher + writer)
OrchestrationThe control layer that decides which agent, tool, or model handles each step“Who does what next” in a multi-agent flow
Copilot / AssistantA human-in-the-loop agent embedded in an app that suggests or assists rather than acting aloneProductivity inside everyday tools

2. Architecture & Reasoning Patterns

How agents are actually wired up under the hood.

TermWhat it isWhat it’s for
RAG (Retrieval-Augmented Generation)Fetching relevant documents or data and feeding them to the model before it answersGrounding answers in your own data; reducing hallucination
GroundingTying model output to authoritative source dataTrust, citations, accuracy
Chain-of-Thought (CoT)Prompting the model to reason step-by-step before answeringBetter results on math, logic, and multi-step problems
ReAct (Reason + Act)A pattern where the model alternates between reasoning and tool-callingThe backbone of most agent loops
Reasoning ModelA model trained to “think” longer before respondingHard problems that need deliberation
Tool Use / Function CallingThe model invoking external functions or APIs in a structured wayHow agents actually do things rather than just describe them
Context EngineeringDeliberately designing everything that goes into the model’s context windowThe evolution of “prompt engineering” for agents

3. Models & Foundations

The engines that power everything above.

TermWhat it isWhat it’s for
LLM (Large Language Model)A model trained on massive text to generate languageThe core engine of most GenAI
SLM (Small Language Model)A compact, cheaper, faster modelOn-device, low-latency, cost-sensitive tasks
Foundation ModelA large pre-trained model reusable across many tasksThe base you build or fine-tune on
Frontier ModelThe newest, most capable models at the cutting edgeBenchmark-leading and demanding workloads
MultimodalHandles a mix of text, images, audio, and videoDocument understanding, vision tasks, voice
GenAI (Generative AI)Umbrella term for AI that creates contentThe whole category

4. Data & Memory

Where agents get their knowledge — and how they remember.

TermWhat it isWhat it’s for
EmbeddingsNumeric vector representations of text or data that capture meaningSemantic search, similarity, RAG retrieval
Vector DatabaseA store optimized for embeddingsFast semantic lookup that powers RAG
Knowledge SourceThe grounding data an agent draws onAccurate, domain-specific answers
MemoryPersistence of context across turns or sessionsPersonalization and long-running agents

5. Protocols & Interoperability

The newest layer — standards that let agents and tools talk to each other without custom glue code.

TermWhat it isWhat it’s for
MCP (Model Context Protocol)An open standard introduced in late 2024 for connecting AI models to external tools and data sourcesPlug-and-play connectors — often described as “USB-C for AI tools”
A2A (Agent2Agent)An open protocol introduced in 2025 (now under the Linux Foundation) for agents to discover, message, and delegate to one anotherCross-vendor collaboration between agents
Connector / PluginA packaged integration giving an agent access to a specific systemExtending agents to enterprise systems
MCP vs. A2A in one line: MCP connects an agent to its tools and data; A2A connects an agent to other agents. They’re complementary, not competing.

6. Build & Tuning

The practical side — making, shaping, and trusting models.

TermWhat it isWhat it’s for
Fine-tuningFurther training a model on your own dataSpecialized tone, domain, or behavior
Prompt EngineeringCrafting effective instructionsSteering model output
InferenceRunning the model to produce output (as opposed to training it)The “cost per use” side of the equation
GuardrailsSafety and policy filters around inputs and outputsPreventing harmful or off-policy responses
Evals (Evaluations)Systematic testing of model or agent qualityMeasuring accuracy and catching regressions
HallucinationConfident but wrong or fabricated outputThe core problem that grounding and RAG try to solve

How It All Fits Together

If you read the tables top to bottom, a rough mental model emerges:

  1. A foundation model (an LLM) provides the raw intelligence.
  2. Wrap it in a reasoning pattern like ReAct, give it tool use, and you have an agent.
  3. Ground it with RAG over a vector database so it answers from real data.
  4. Connect it to systems via MCP, and to other agents via A2A.
  5. Coordinate several of them through orchestration to build a multi-agent system.
  6. Keep it honest with guardrails and measure it with evals.

That progression — from a single model to a coordinated, tool-using, self-correcting system — is essentially what people mean when they say agentic AI.


Know a buzzword that should be on this list? The vocabulary keeps growing — this glossary will too.

No comments:

Post a Comment

Featured Post

The AI Buzzword Glossary: From Agents to Agentic AI

The AI Buzzword Glossary: From Agents to Agentic AI Every week brings a fresh wave of AI terminology. “Agentic AI,” “RAG,” “MCP,” “co...

Popular posts