Operations managers need a practicum built around running the workspace, not clearing individual cases. For two weeks the manager makes the live operational calls, such as rerouting held work, responding to a certificate drop and approving limit changes, while an assessor marks each one. Pass at 95 correct in every 100, with no missed red line.
Why is an operations manager's practicum different?
A reviewer's practicum is a stream of held cases. An operations manager's is a stream of decisions about the system that produces those cases: whether the queue is staffed, whether an agent's limits should change, what to do when an Attest run clears no depth and an agent drops to depth 1 that afternoon. The work is less frequent and the consequences are wider.
So the unit of assessment changes. Each operational decision the manager makes, with the reason they recorded, is one mark. The head of operations owns the practicum; the assessor should be someone who has run the same workspace and is not in the manager's reporting line. The guide to AI literacy and certification for people who work with agents covers why practical sign-off matters as much as the written exam.
What does the manager do over the two weeks?
- Person (head of operations): hand over the live workspace. The manager runs the daily rhythm: morning queue review, midday check on response targets, end-of-day handover.
- Agent, depth 1: reports queue state. Agents in the workspace show what they are holding, why, and for how long. The manager reads this, not a weekly report.
- Person (manager): reroutes and staffs. Moves held work to the right reviewers, calls in cover when a hold reason is breaching its response target.
- Person (manager): decides on limit-change requests. Teams ask for an agent's limit to rise or fall. The manager approves within their authority and refers anything beyond it.
- Agent: drops to depth 1 when a run clears no depth or a certificate lapses. This happens the same day. The manager decides how to absorb the work it can no longer do and who requests recertification.
- Person (manager): handles one incident end to end. If no live incident arises, the assessor runs a replayed one in a training workspace, recorded as replayed.
- Person (manager): reviews usage and billing for the period. Spot capabilities consuming more than expected and explain why.
- Person (assessor): marks each decision and runs a midpoint review.
- Person (assessor with head of operations): sign off or extend.
| Step | The agent does | Depth | Stays with a person |
|---|---|---|---|
| Daily rhythm | Reports holds, reasons and ages | 1 | Deciding what needs attention first |
| Rerouting | Keeps holding until a named person acts | 2 to 4 | Who reviews what, and when |
| Limit changes | Works within whatever limits apply | 3 | Approving or refusing the change |
| Certificate drop | Falls to depth 1 the same day | 1 | Absorbing the work and requesting recertification |
| Incident | Files every verdict as it happens | 1 | Containment, communication and recovery |
| Usage review | Records usage by capability | 1 | Explaining and acting on the pattern |
What are the manager's red lines?
Four, written down before day one:
- Raising an agent's limits without evidence from Attest that it can work at the new level.
- Letting a hold involving a red-line case, such as a payment to a changed bank account, sit past its response target without escalation.
- Working around a certificate drop, for example by having people approve in bulk what the agent can no longer do, without checking each item.
- Approving a change outside the manager's own authority.
Any one of these ends the practicum.
Which controls keep the assessment fair, and what evidence do they leave?
- Every operational decision is filed. Limit changes, reroutes and incident actions are verdicts in the append-only record, with the manager's name. The assessor marks from that record.
- Reasons are mandatory. A limit change with no recorded reason counts as wrong even if it turns out well.
- Replayed incidents are labelled. The record shows which events were live and which were replayed, so nobody later reads a practice incident as a real one.
- The assessor sees the same screens. Marking from a summary the manager wrote is not assessment.
What should be measured?
- Decision accuracy against the 95 in 100 bar. Expect fewer decisions than a reviewer makes, often a few hundred over the period; each one matters more.
- Response targets met per hold reason, under the manager's staffing.
- Time from certificate drop to a workable plan.
- Limit-change quality: the share of changes backed by Attest evidence and a clear reason.
- Handover completeness: whether the next shift could act on the manager's notes without asking.
The practicum log template for operations managers gives columns for each, and pairs with the risk managers' log where the two roles share a workspace.
Where does it usually go wrong?
Nothing happens. A calm fortnight produces no incidents and no limit requests. Seed them: a replayed certificate failure, a request from a team lead to raise a refund limit, a backlog after a public holiday.
The manager reviews cases instead of running the system. Capable managers drift into clearing held items themselves. That keeps the queue moving and hides whether they can staff it.
Assessment by outcome. A manager who skipped an escalation and got lucky has still made a wrong decision. Mark the decision and the reason, not the result.
Two related playbooks show the same method elsewhere: the practicum for risk managers focuses on limits and risk acceptance, and the one for property managers on a single operation with many small decisions.
Where the Academy fits
The Operator track is written for the people who run a live workspace every day, and operations managers are named among them. Its modules on reading the workspace, when something goes wrong, usage and billing, and running the rhythm map to the steps above. The practicum is two weeks of real decisions signed off by a Pryme assessor, and the certificate is valid for 12 months.
Questions people ask
How many decisions does an operations manager need for a fair score?
Enough that one bad call does not decide the outcome on its own, unless it is a red line. If the fortnight produces too few, extend it or add replayed events rather than lowering the bar.
Should the manager also review held cases?
Some, so they stay close to the work. But the practicum should be weighted towards operational decisions, which is what the role adds.
Who assesses the head of operations?
Someone external to the line, often a peer from another business unit or a certified assessor. The principle is the same at every level: nobody marks their own manager's work.