An assessor is a qualified person who judges, from evidence, whether a learner has met a defined standard, and signs off that judgement. In training for AI agent oversight, the assessor looks at a person's real review decisions and decides whether they can be trusted to approve an agent's work without supervision.
Why does the assessor matter?
A credential is only as strong as the judgement behind it. A score from an automated quiz shows what someone recalled. An assessor can look at how they reasoned on actual cases: whether they rejected a payment backed only by an emailed invoice, whether they escalated something outside their authority, whether their edits were precise. That is the argument in proof of real work beats a multiple-choice score.
The assessor also carries accountability. Their name sits behind the sign-off, so a client or supervisor asking "who decided this person was competent?" gets an answer.
What does an assessor do in practice?
For someone learning to oversee agents, the assessor typically:
- agrees the standard and the evidence needed before the assessment starts;
- reviews a sample of the learner's real decisions, including the hard ones;
- checks for red lines, the errors that fail the assessment on their own regardless of overall accuracy;
- asks the learner to explain decisions that look wrong or lucky;
- records the outcome and the reasons, and signs off or sets a further period of practice.
An assessor should not be the learner's line manager when that can be avoided, and should never assess their own work.
What is an example?
A contact centre lead completes two weeks of reviewing an agent's held refund drafts. The assessor finds the overall accuracy is strong, but one refund was approved to a changed bank account with no verified evidence. That is a red line. The assessor does not sign off, sets a further practice period, and records why. The lead's accuracy was good; their judgement on the one decision that mattered most was not yet reliable.
Outsourcers and contact centres use this kind of sign-off as proof for clients, as covered in how contact centres prove staff are trained to oversee agents and how outsourcing providers prove it.
How is an assessor different from a trainer?
A trainer teaches; an assessor judges. One person can do both in different settings, but not for the same learner on the same standard, because the person who taught has a stake in the result. An assessor also differs from an invigilator or proctor, who supervises an exam to make sure the rules are followed but does not judge the answers.
Assessors themselves are qualified by the body issuing the credential. Whether that body is formally recognised is a separate question, covered in what accreditation is.
Which terms are related?
- Practicum: the period of real work the assessor reviews.
- Board review: a panel of assessors examining a candidate together.
- Evidence of competence: what the assessor's sign-off produces.
The full topic is in AI literacy and certification for people who work with agents.
Where the Academy fits
In Pryme Intelligence Academy, the Certified Pryme Operator practicum is two weeks of real review decisions, at least 95 in every 100 correct with no missed red line, signed off by a Pryme assessor. Certified Pryme Architect ends with a board review before two Pryme assessors.
Questions people ask
Can a line manager be the assessor?
It is better if not. A manager has an interest in the result and may be judging their own team's process. Use an independent assessor, or at least one from outside the learner's reporting line.
What should an assessor record?
The standard, the evidence reviewed, any red lines, the outcome, the reasons and their own name. That record is what makes the sign-off defensible later.