Reviewer onboarding checklist for operations team

Operations reviewers handle whatever the agent cannot finish: exceptions, breaks, items outside bounds and the work that floods back when a certificate lapses. Onboard them with access and red lines, shadowing on the exceptions queue, reviewer training and exam, a practicum that includes a lapse day, a drill, and sign-off that puts them on the rota.

What is the operations reviewer's job with an agent?

In an operations team, the agent processes routine work at depth 3 within bounds: matching, updating records, routing requests, sending standard confirmations. It holds what does not fit. Those exceptions are the reviewer's daily work, and they vary: a record that will not match, a request outside the agent's scope, a value just over a limit, a customer instruction from an unusual channel.

Two features set operations apart. The first is volume variation. When the agent's certificate lapses, or a certification run that clears no depth drops it to depth 1 the same day, routine work returns to people at once. An operations reviewer must know how to run the queue manually without lowering the standard. The second is cover. Operations teams run rotas, and a held action that waits for an absent approver can breach a service level. Reviewers need to know who covers them, and that cover must be certified too.

How should the checklist be run?

The operations manager owns it. Run it in order and do not add the reviewer to the approval rota until stage 5 is signed. Once they are signed off, plan their place on the rota as in how to build a reviewer rota for an onboarding team, a method that works for any operations queue. The general idea of onboarding applies, and the wider training order is in the guide to training teams for AI agents.

The checklist

Stage 1: set-up - [ ] Read-only access to the exceptions queue and the record - [ ] Agent capabilities read, with depth, for every process the team covers - [ ] Red lines signed: never approve an exception outside your rights to clear a backlog; never act on an instruction from an unverified channel; never approve an item you created - [ ] Buddy named

Stage 2: shadowing (week 1) - [ ] Two days on the exceptions queue - [ ] Seen an exception refused and returned with a reason - [ ] Walked through the lapse procedure: what falls back to people, who decides priorities

Stage 3: training (weeks 1 to 5) - [ ] Shared foundation complete, test passed - [ ] Reviewer modules complete, written exam passed

Stage 4: practicum (weeks 6 and 7) - [ ] Two weeks of real held exceptions, each confirmed by the buddy - [ ] Coverage: a mismatch, an over-limit item, an unusual-channel instruction and at least one day run with the agent at reduced depth - [ ] Incident drill completed, following a format like an AI agent incident drill for the onboarding team

Stage 5: sign-off - [ ] Assessor sign-off; no missed red line - [ ] Approval rights by process and value - [ ] Added to the rota with a named, certified cover - [ ] Renewal date set

Worked example

An illustrative insurance operations team of 20 brings in Sample Reviewer BB. The agent updates policy records, issues standard documents and processes mid-term changes within bounds; it holds changes that alter premium by more than 100 and any request arriving by post or unrecognised email.

During shadowing, Sample Reviewer BB sees the buddy refuse a held address change that came from an email address not on the policy, and return it to the agent with a note to request confirmation from the policyholder. She also walks through the lapse procedure, which ranks work by customer impact if the agent drops to depth 1.

Her practicum includes a real reduced-depth day: a certification run clears no depth on a Tuesday and the agent drops to depth 1. The team processes 140 mid-term changes by hand; she handles 30, with her buddy checking each. She is signed off in week 7, added to the early shift rota, and given Sample Reviewer CC as certified cover.

How does general operations differ from specialist operations?

Specialist operations teams carry extra red lines. Payments operations has cut-offs and irreversible transfers, covered in the reviewer onboarding checklist for the payments operations team. Legal operations has privilege and notice dates, covered in the reviewer onboarding checklist for the legal operations team. Use this checklist as the base and add theirs where a reviewer covers both.