AFI / PROPOSED OPERATING MODELA workforce of agents.
A chain of accountability.
Routine work scales through specialized agents. A small human leadership team owns policy, decisions, exceptions and recovery.
HUMAN ACCOUNTABILITYAGENT WORKINDEPENDENT CHECK
HUMAN GOVERNANCEBoard & oversight
Strategy · risk appetite · accountability
HUMAN LEADERSHIPCEO / accountable executive
Owns the institution and coordinates five operating functions.
AGENT COORDINATION LAYERAFI institution orchestrator
Routes work, tracks state and evidence, calls specialist agents and escalates exceptions. Human leaders own each function below; the orchestrator does not replace them.
Member experience
HUMAN OWNERMember experience leadComplex service, complaints, accessibility and recovery.
Onboarding agent
Assembles the application and consent checklist.
Member concierge
Routes goals and explains the next step.
Financial planning assistant
Prepares member-directed budgets and savings proposals.
Dispute intake agent
Collects facts, evidence and deadlines.
Recovery coordinator
Routes access failures to a human; cannot override identity checks.
Lending
HUMAN OWNERCredit leadOwns credit policy, exceptional decisions and lending outcomes.
Application assembler
Packages income, purpose and evidence.
Document reviewer
Checks completeness, flags inconsistencies.
Policy evaluation agent
Runs approved checks; flags out-of-policy requests.
Offer explainer
Explains terms and compares requested scenarios.
Loan servicing agent
Tracks payments and prepares support or hardship cases.
Risk & compliance
HUMAN OWNERRisk / compliance leadOwns compliance policy, suspicious cases, fraud response and model risk.
Fraud monitoring agent
Detects anomalies and creates investigation cases.
AML / KYC triage agent
Collects evidence and prioritizes screening exceptions.
Consent monitor
Checks scoped access, limits and revocation.
Model evaluation agent
Tests drift, failures and unsafe behaviors.
Compliance evidence agent
Maintains review packages and policy traceability.
Operations & finance
HUMAN OWNEROperations / finance leadOwns the ledger, settlement, liquidity and vendor relationships.
Payment operations agent
Prepares validated actions after approval.
Reconciliation agent
Matches transactions and flags ledger differences.
Liquidity monitor
Produces forecasts and threshold alerts.
Finance reporting agent
Drafts reporting from approved records.
Vendor coordinator
Tracks service incidents, cost and obligations.
Technology & security
HUMAN OWNERTechnology / security leadOwns access, architecture, incidents and releases.
Agent gateway
Verifies identity, capability scope and request shape.
Security monitor
Detects suspicious access and routes incidents.
Reliability agent
Monitors uptime, queues and failed workflows.
Engineering assistant
Prepares code and tests for reviewed releases.
Knowledge curator
Keeps approved policies and documentation current.
INDEPENDENT ASSURANCEIndependent audit & specialist review
Reports to governance, separate from delivery. Legal, privacy and specialist expertise can be shared or external. Monitoring agents prepare evidence; people independently challenge the result.
Agent-first, with a human owner.
One human may hold several roles where appropriate. This is a role map, not a validated staffing minimum. Coverage, independence and staffing depend on the institution’s charter, jurisdiction, products, volume and risk.
The action boundary
- Member agent prepares the request.
- AFI agents validate scope, evidence and policy.
- Member approves the exact action.
- Institution controls execute and record it.
- Exceptions reach the accountable human owner.
Concept organization only. No autonomous hiring, credit decision, suspicious-activity filing or real transaction is performed by this site. Agents must operate inside institution-controlled policies and authorization.
Back to AFI