How to onboard a new collections agent to an AI agent workspace

Onboarding a new collections agent where AI agents already send reminders and offer payment plans takes two weeks: one learning what the AI agent may agree to and what it must hand over, one handling held accounts with a coach checking each outcome. Vulnerability and disputes are the two lines a new person must hold before working alone.

What is the AI agent doing in collections?

A word on terms first: in collections, "agent" usually means the person on the phone. Here the new collections agent is the person; the AI agent is the software working beside them.

In a collections team that runs AI agents, the AI agent typically sends reminder messages on a set schedule, answers simple balance and due-date questions, offers payment plans within the creditor's rules (depth 3 for plans inside set limits), and drafts letters and plan proposals outside those limits for a person (depth 2). It must hand over any account where the customer mentions hardship, illness, bereavement or a dispute. It never threatens, never contacts at prohibited times, and never discusses the debt with third parties.

The new collections agent then deals with the held accounts: plans above limits, disputes, vulnerable customers and complaints. These are the conversations that matter most to customers and regulators. The collections team leader owns onboarding; a certified senior colleague coaches.

What happens over the two weeks?

1. Day 1. Person (team leader): walks through the AI agent's bounds, especially plan limits and the hand-over triggers.
2. Day 1 to 2. Person (new collections agent): learns what the AI agent can see. The account balance from the ledger is sealed. A customer's statement of income is declared. A letter from a debt adviser is signed by an identifiable party.
3. Day 2 to 3. Agent, depth 2: drafts plan proposals and letters on held accounts, worked by the coach while the new person listens to calls and reads messages.
4. Day 4 to 5. Person (new collections agent): practises in a sandbox workspace on past held accounts, including a hidden vulnerability case and a hidden dispute.
5. Week 2. Person (new collections agent): handles live held accounts, with the coach checking each outcome before it is confirmed to the customer.
6. Week 2. Person (new collections agent): records reasons whenever they change the AI agent's proposal.
7. End of week 2. Person (team leader): signs off solo handling for standard held accounts; vulnerable and disputed accounts may stay supervised longer.

Which lines must a new person hold?

Vulnerability. Customers rarely say "I am vulnerable". They say they have lost their job, a partner died, they are in hospital. The AI agent is bounded to hand these over; the new collections agent must recognise them in conversation too and adjust: pause collection activity where policy says, offer appropriate support, record what was agreed.

Disputes. If a customer says the debt is wrong, collection activity on the disputed amount stops until the dispute is resolved. A new person must not keep pressing because the AI agent's proposal is still on screen.

Plan affordability. A plan the customer cannot keep helps nobody. The new person should check affordability, not just accept the first figure the customer offers or the AI agent proposes.

How do they correct the AI agent?

With a reason each time: plan too high for stated income, vulnerability missed, dispute missed, wrong balance source, or person's judgement. Reasons let the agent owner change the AI agent's brief, sources or bounds where the error is real. The AI agent training plan template for collections agents has a module on this.

Which controls and evidence should onboarding leave?

• Checked outcomes in week two. Evidence: both people named on each agreed plan.
• Sandbox cases including vulnerability and dispute. Evidence: results recorded.
• Staged rights. Evidence: dated access for standard accounts, later for vulnerable ones.
• Reasons on every change. Evidence: reason codes by person.

What should the team leader measure?

Vulnerability cases recognised. Disputes correctly paused. Plans agreed that the customer kept for the first few months. Complaints on accounts the new person handled. And the quality of correction reasons.

Where does it usually go wrong?

Following the AI agent's script. The proposal on screen feels authoritative, and a new person reads it out even when the customer has just described hardship.

Treating sign-off as complete. Vulnerable customers need experience to handle well; keep them supervised beyond two weeks if needed.

Not knowing who to ask. New collections agents should know the escalation route for complaints and for unusual AI agent behaviour, which is usually the operations manager; the plan for a new operations manager covers their side. Support teams onboard new people with similar staged rights, as the plan for a new customer support agent shows, and the overall approach is in training teams for AI agents.

Where the Academy fits

Pryme Intelligence Academy's Operator track teaches how a governed agent decides and how to review its held work, with module tests set in the learner's own industry and role. A collections business can choose to make the right to agree plans outside limits depend on a current Certified Pryme Operator certificate, which is valid 12 months.