How to set an approver rule for the operations team

Operations approvals fall into four groups. Team members approve the agent's exceptions to routine work. A team lead approves changes to standing data that affect money. Bulk actions need the operations manager plus a second approver. House rule changes belong to the operations manager alone. Every right depends on a current certificate.

What does an operations approver rule cover, and who owns it?

Operations teams run the queues behind a business: requests, updates, documents, exceptions, service levels. Agents take on the routine part of those queues at depth 3, within house rules. The approver rule handles everything else: exceptions the agent holds, changes that affect money, and actions that touch many records at once.

The operations manager owns the rule, with risk signing off. The agent owner configures it. Our explainer on what a house rule is sets out how house rules bound the agent, and our guide to AI agent skills by role covers what operations staff need to know.

What are the steps?

1. Operations manager: lists the agent's actions by depth. Depth 1: reads queues, records and service-level timers. Depth 2: drafts customer and supplier communications for non-standard cases. Depth 3: processes routine requests within house rules, such as verified address changes and document requests. Depth 4: changes standing data that affects payments, applies bulk updates, closes records.
2. Operations manager: assigns exceptions to team members. Anything the agent holds because it falls outside a house rule.
3. Operations manager: assigns money-affecting standing data to a team lead, after verification by a channel on file.
4. Operations manager: assigns bulk actions to themselves plus a second approver, with a sample check before release.
5. Operations manager: keeps house rule changes to themselves, with risk told of any change that widens what the agent may do.
6. Operations manager: ties each role to a current certificate. Agent owner configures and files.

Why do bulk actions need their own line in the rule?

A single wrong update is a mistake. A wrong bulk update is an incident. When an agent prepares a change to thousands of records, such as a new fee applied to a product line or a status change after a migration, one approval releases all of it. The approver cannot read every record, so the rule must say what they read instead.

The agent should present the bulk action with a count, a description of the selection logic, a random sample of before-and-after records, and any records that failed validation. The first approver checks the selection; the second checks the sample. If either cannot explain the selection in a sentence, it is not released.

House rules are the other operations-specific point. They decide what the agent does without asking. Changing one is like changing a delegated authority for every request that follows, so it sits with the operations manager and is recorded.

Which controls make the rule work, and what evidence do they leave?

Exceptions owned by named team members. Evidence: approver names on held exceptions.

Verification before money-affecting changes. Evidence: verification channel and result on each change.

Two approvers and a sample on bulk actions. Evidence: the sample shown to approvers and both names, as set out in the approver authority matrix for the operations team.

House rule changes recorded. Evidence: each change with its reason, date and the manager's name.

Rights lapse with certificates. Evidence: certificate status on each approval date.

What should you measure?

• Exceptions per hundred requests, by house rule. A high rate may mean the rule needs revisiting.
• Bulk actions reversed. The target is none.
• Standing data changes without verification. The target is none.
• Service-level performance on held items, which shows whether approvers are keeping up.

Where does it usually go wrong?

House rules multiply. Every exception becomes a new house rule, and soon nobody can say what the agent will do. Prune them quarterly, and record why each one exists.

Bulk actions are released at the end of the day. Tired approvers release large changes before leaving. Set a cut-off time.

Team leads approve their own exceptions. A team lead who also works the queue holds an exception, then approves it from the lead's seat. The rule should say that nobody approves an exception they raised or worked.

Cover is planned for averages. Exceptions spike after a product change or a system incident. The agent holds more, approvers fall behind, and service levels slip on exactly the cases that need care. Name surge approvers in advance, all with current certificates.

Payments and legal operations are treated as ordinary operations. They need their own rules; compare the approver rule for the payments operations team, its approver authority matrix, and the approver rule for the legal operations team.

Where the Academy fits

Pryme Intelligence Academy's Operator track is aimed at operations managers and team leads, with modules on reading the workspace, running the rhythm and when something goes wrong. Certified Pryme Operator requires a written exam of 40 questions with a pass at 32 or more, plus a practicum signed off by a Pryme assessor. The admin view shows each approver's certificate and expiry.