Chargeback Agent
Meters usage of every model, dataset and agent against entitlements, applies the published rate card and drafts each business unit's monthly chargeback to its cost centre, in MYR. Variances are flagged for Group Finance to clear; the journal is posted by a person.
An agent gathers the evidence, applies the rule, scores the case and recommends an outcome. A named human decides and the decision is recorded against them.
- Acts for
- Marketplace Operator
- Action class
- advisory
- Decisions / 30d
- 7
- Overturned by a human
- 0%
Processes covered
- Usage metering
- Chargeback rate card
- Chargeback posting to cost centres
Memory
Retains reconciled periods for 7 years under financial record rules.
Safe state
Produces no statements and alerts Group Finance when the metering ledger and the entitlement registry disagree by more than 0.5%.
Reviewer time per case: 20 min with this agent, 240 without.
Ownership and sensitivity
Who approves access
- Owner, Platform Engineering, Hafiz Othman
Business-sensitive. Models and data products scoped to named business units.
Entitlement per business unit, approved by the owner; conditions attach.
- Aggregated only
- No row-level customer data; aggregates with small-cell suppression.
Tools
- Usage metering ledger
- Entitlement registry
- Chargeback rate card
- Statement drafter
- Variance flagger
Data access scope
- Read-only: metered calls, compute usage, entitlements and rate card
- Write: draft chargeback statements and variance flags only
- No access to the general ledger
Guardrails
- Cannot post a journal; Group Finance does
- Any variance above 2% of a statement blocks the draft until cleared
- Usage is reconciled per entitlement, so a call outside an entitlement is a flag, not a charge
Human-in-the-loop
- Group Finance clears variances and posts the chargeback journal
Orchestration
Every tool call is scoped by the declared data access above; the orchestrator cannot reach systems outside it. Owning business unit: Group Data & AI.
Live demo run
Watch the agent execute a real scenario step by step, every tool call, validation, and human checkpoint is traced and auditable. Typical run: ~5.5s.
Used in workflows
Retiring this asset would require these chains to be re-pointed first.
Audit
Updated 2026-07-01. Run traces retained 24 months for audit under the platform governance policy.
Community · 0 threads
Questions, findings and requests from the business units that use this asset. Owners reply here; threads with upvotes surface to the owning team's inbox.