What is a readiness assessment?

A readiness assessment is a structured test of whether a team has the knowledge and habits to put an AI agent to work safely, carried out before the agent goes live. It measures people, not software: whether they understand how the agent decides, what a held action asks of them, and where their own approval is the last line.

Why does a readiness assessment matter?

Most failed agent deployments do not fail on the model. They fail because the people around it approve what they should question, ignore what they should escalate, or do not know that a lapsed certificate has dropped the agent to read-only. A readiness assessment finds those gaps while they are cheap to close.

It also tells you where to start. A team that scores well on source provenance but poorly on limits may be ready for an agent that drafts letters at depth 2, and not yet for one that prepares payments for approval.

What does it measure?

A useful assessment tests concepts that map to real decisions. A common set:

• How a governed agent reaches a verdict, and which bound decided.
• The difference between held, refused and withheld.
• Which sources can support which action.
• What the reviewer's approval adds, and what it does not.
• What happens when an agent's certificate lapses or a run fails.

The output is a report by concept, not a single score. Our article on what a readiness assessment for AI agents measures goes through each area, and the readiness assessment report template shows how to lay out the findings.

An example

A mid-sized bank plans an agent that drafts responses to account-change requests. Twenty people across operations and the contact centre take a short test set in banking. The report shows strong scores on verdicts and weak scores on provenance: half the group would accept an emailed letter as proof of a new address. The bank spends two weeks on provenance before go-live and starts the agent at depth 2. The guide to running a readiness assessment in a bank covers the steps; a fintech version adjusts them for smaller teams.

How is it different from operational readiness?

Operational readiness asks whether the process, systems and support are in place: connections tested, limits set, escalation routes staffed, rollback planned. A readiness assessment asks whether the people are ready. You need both. A perfectly configured agent with an untrained review team is not ready, and neither is a trained team with no escalation route. See what operational readiness covers.

It also differs from certification. An assessment is diagnostic and carries no rights. A certificate is a decision about one person and can be tied to what they may approve.

Which terms are related?

• Training needs analysis: the wider study of what each role must learn, which an assessment feeds.
• Skills gap: the distance between what the assessment finds and what the role needs.
• Operational readiness: the system and process side of go-live.

For the whole topic, read training teams for AI agents.

Where the Academy fits

Pryme Intelligence Academy offers a free readiness assessment for teams of up to 25 people. The team takes the Foundations test set in its own industry. The report shows scores by concept, the gaps, and the first agent to start with, so the training plan and the first deployment are chosen from the same evidence.