AI Certificates

Reference

CCAR-P glossary: the terms the exam conflates

Matthew Hartman · CCAR-P certified, scored 965/1000 · 3 min read

Most of these are not hard words. They are words that sit close enough together that using one for another produces a confident, wrong architecture, which is exactly what the exam tests.

The four that get conflated most

Keeping these apart settles a surprising number of questions on their own.

TermWhat it isWhat it is not
Context windowThe model's active attention space for a single call.Storage. It resets between calls unless your application refills it.
RetrievalExternal knowledge fetched at query time and placed into the window.A replacement for the window. The model still only sees what fits.
Persistent application stateValues your system owns and changes: order status, balances, records.Something the model can reach without a tool call.
MemoryContinuity your application stores and replays across turns.A model capability. There is none between calls.

The most common architecture error in this group is representing live state as retrievable text. Retrieval is for knowledge that was true yesterday and remains true tomorrow. Tool calls are for values a system owns and changes independently.

The seven primitives

Patterns are assemblies of these rather than categories of their own.

PrimitiveIts one job
ToolsLet the model take an action or fetch a result from your code.
MCPExpose a set of tools so several clients reach the same endpoints.
SubagentsHand a scoped sub-task to a separate context, in isolation or in parallel.
HooksFire deterministic code on an event to enforce a rule the model cannot skip.
SkillsPackage a repeatable procedure as a versioned, reusable unit.
Agent teamsCoordinate peer agents that each own part of a goal.
Dynamic workflowsAssemble the steps of a workflow at run time rather than fixing them.

Three layers people answer across

LayerThe question it answersChosen for
Entry pointWhat does a person or system touch?The user and the work
Build-time interfaceWhat do engineers write against?The team and the integration
Delivery routeWhose infrastructure does the request terminate on?Compliance and cloud commitments

Evaluation vocabulary

TermMeaning
Code-based evalA deterministic check, milliseconds, near-zero cost. For anything with one right answer.
Model-based evalA judge model scoring qualities needing interpretation. Costs about as much as the call it grades.
Human-review evalHuman judgement for high-stakes or novel behavior. Does not scale, so it is reserved.
Golden datasetThe inputs an eval runs against, including edge cases and counterexamples.

Guardrail decision points

A control at one point does nothing for the others.

PointRunsUsually
Input screeningBefore the model callModel-based, since injection phrasing cannot be enumerated
Output screeningBefore the response reaches the userModel-based, since policy compliance needs language understanding
Tool-call authorizationBefore any side-effecting actionDeterministic, because authorization must be auditable

If a term on the exam sent you looking and it is not here, tell me in the Q&A on any of the practice courses and I will add it. A word you had to translate under time pressure is exactly what a glossary is for.

Practice this

A full-length 63-question CCAR-P form. Free, no signup.

Start form 1 · study mode

The full set is on Udemy.