Agentic commerce
Score what arrives. Let the good agents in.
Decide which agent payments to accept and set limits for your own agents. One console, one API, one shared record.
Inbound · someone else's agent
Risk Scoring
A decision before authorization: approve, ask the customer, refer or decline, with the reason codes behind it.
- Six factor families, two of them hard gates
- No card numbers, CVV or usable cryptograms
- Advisory by default; you choose when it blocks
Outbound · your own agents
Agents you operate
Register every agent you run and grant authority with a ceiling in currency.
- Capabilities carry limits in money, not API scopes
- Every rule resolves to off, shadow or enforce
- Missions hold a budget many agents draw against
Sharp Trust · One console + one API
One shared record.
Connect the decision before a payment with what happens after it.
Decisions + reasons
What was approved, challenged, referred or declined, and why.
Mandates + budgets
Who gave the agent authority, and how much it could spend.
Refunds + disputes
What happened after the payment, linked back to the decision.
Payment outcomes give future risk decisions more context. Your agents and your scoring draw on the same history.