PrymeIntelligence

AI agent skills by role·Template

Approver authority matrix for lending team

An approver authority matrix for lending teams: who approves agent-prepared loan decisions, exceptions, disbursements, decline letters and restructures.

A lending approver authority matrix shows, for each lending action an AI agent prepares, who must approve it and up to what amount. Agents collect documents, verify evidence and draft decision packs; loan officers and credit managers approve applications within mandate, disbursements are held at depth 4, and every decline letter is checked for specific reasons before it goes.

How should approval work when agents prepare loans?

Lending decisions about individuals are among the most closely watched uses of automation. Under the EU AI Act, creditworthiness assessment and credit scoring of natural persons are high-risk. In the US, ECOA and Regulation B require specific reasons when credit is denied, including when a model is used. Whatever the market, a lender must be able to show that a person decided and could explain why.

That makes the agent's role clear. It reads applications and bureau data at depth 1, verifies documents and drafts decision packs at depth 2, and prepares disbursements, which are depth 4 and held. The matrix assigns each decision to an approver by amount and risk, sets the second approver, and states what the approver must check rather than accept from the pack. The logic is that of any authority matrix.

How do you fill it in?

  1. Start from your lending mandate: who may approve what, by product and amount.
  2. For each action, write the agent's part and depth.
  3. Keep the approver limits from the mandate; agents do not change them.
  4. Add a second approver where the decision is an exception or large.
  5. Add the check the approver must perform personally.
  6. Record the certificate required, if approval rights are tied to certification.

The template

Lending action Agent's part and depth Approver Limit Second approver when Approver must check Certificate
Approve application within policy
Approve with policy exception
Decline application
Waive a required document
Release disbursement
Agree restructure or payment holiday
Waive fees
Change interest rate

A worked example

A hypothetical consumer lender: twelve loan officers, three credit managers and a head of lending. Values are illustrative.

Action Agent's part Approver Limit Second approver when Must check Certificate
Approve within policy Drafts pack (2) Loan officer 25,000 Above 25,000, credit manager Income evidence provenance Operator
Approve with exception Drafts pack (2) Credit manager 50,000 Always, head of lending Exception rationale Operator
Decline Drafts letter with reasons (2) Loan officer All Applicant is an existing customer, credit manager Each reason traces to a fact Operator
Waive document Summarises gap (1) Credit manager All None Alternative evidence Operator
Disbursement Prepares payment (4, held) Loan officer 25,000 Above 25,000, credit manager Account in applicant's name Operator
Restructure Drafts options (2) Credit manager All Third restructure, head of lending Affordability and vulnerability Operator
Fee waiver Drafts (2) Loan officer 250 Above 250, credit manager Reason recorded Foundations
Rate change Summarises (1) Head of lending All Always, risk Pricing policy Operator

The decline row has the sharpest check. The agent drafts the letter and its reasons. The officer confirms that each reason points to a fact in the file, and rewrites any reason that is generic. A decline that says "does not meet our lending criteria" without saying which one fails the applicant and, in some markets, the law.

Disbursement is held even for small loans, and the officer checks that the receiving account is in the applicant's name. An account change requested by email after approval is external and cannot authorise the payment; the agent will refuse it, and the officer should verify any genuine change through a known channel.

Document waivers sit at depth 1: the agent may describe what is missing, but it does not argue for the waiver. The credit manager forms their own view.

System access for the lending agent is approved under the approver authority matrix for the IT team, and operational exceptions such as missed service levels under the approver authority matrix for the operations team. Detailed thresholds belong in the approver rule for the lending team.

Where the Academy fits

Pryme Intelligence Academy's Foundations track gives everyone in lending the shared vocabulary, which supports the AI literacy expected of staff. Loan officers and credit managers who approve held decisions take the Operator track, with tests set in lending. A lender can make approval rights depend on a current certificate, depth by depth.

Questions people ask

Can the agent approve small loans on its own?

Some lenders may consider it for very small, policy-standard loans once the agent has earned that depth on their own cases. Where individuals are affected and rules require explainable decisions, many keep all approvals with people. Who needs which skill is set out in AI agent skills by role.

Who approves when an officer's certificate lapses?

Another certified officer. The lapsed officer can still prepare and review, but not approve, until they renew.

Should credit managers see every decline?

Not every one, but a weekly sample, and every decline for an existing customer. Patterns in declines are an early sign of drift in the agent's drafts.

Train the people who run your agents.

Pryme Intelligence Academy: free courses on the real product, audio lessons, labs in a free workspace, and certificates anyone can check on VerifyDoc.ai. Teams get seats, an admin view and a free readiness assessment.