Servicing Plane overview
The servicing control plane is the operating layer for loan and merchant-advance servicing — coordinating cases, work, compliance, people, and AI agents around a system of record it does not own.
The servicing control plane is where the operational work of servicing happens. It coordinates cases, tasks, and channels across your human agents and AI agents, gates every borrower contact and account change through compliance, and drives each matter to an outcome — all around a system of record it deliberately does not own. Consumer installment, BNPL, and merchant-advance servicing run on the same model.
A control plane, not a CRM
A CRM records what happened. The servicing control plane tracks the outcome you are trying to reach: it holds the operational state you want, submits approved requests to your system of record, verifies the core actually reached that state, and creates work when reality does not match. Divergence surfaces as drift and becomes prioritized work — never hidden behind a completed task.
Who owns what
The split is deliberate and never blurred. The system of record owns financial truth; the servicing plane owns the operational work around it.
| The system of record owns | The servicing plane owns |
|---|---|
| Balances and repayment schedules | Cases, tasks, and queues |
| Payment allocation and transactions | Interactions across phone, email, and chat |
| Accounting and delinquency calculation | Compliance decisions and approvals |
| Charge-off, payoff, and reversal | Action-request lifecycle and evidence |
| Autopay and restriction facts | Promises, AI governance, and audit |
How work moves
A case is opened from a signal or by hand and carries the matter to closure. Work reaches an eligible worker — human or AI — through smart queues. Before any borrower contact or account change, the compliance engine returns a structured decision, and account-changing work runs through the action request gateway under maker-checker approval. Every decision, approval, and override lands in the evidence graph.
Automation comes in two forms
The plane draws a hard line between two kinds of automation. Deterministic, policy-gated automation — reminders, notices, and routing — is templated and rule-bound, and each instance still passes a per-recipient compliance check. AI autonomy is different: an agent exercises judgment, bounded by its work packet and the autonomy ladder. You can run substantial deterministic automation well before AI autonomy rises.
Explore the servicing guides
- Case model & lifecycle — how a matter is modeled, tracked, and closed.
- Smart queues & routing — how work reaches the right worker.
- Compliance engine — the decision gate in front of every contact and action.
- AI agent workforce — AI agents as first-class governed workers.
- Reconciliation & drift — keeping operational and authoritative state in agreement.