Fraud & Risk Add-on FAQ
The questions a risk or security team should ask about any fraud product, answered plainly for the Fraud & Risk Add-on for Agentic Payments.
The Fraud & Risk Add-on for Agentic Payments is a separately priced add-on to AxonFlow Enterprise. It is not included in the base Enterprise license, the Evaluation tier, or the source-available Community edition. It requires platform v9.18.0 or later and is currently available through the early-access program.
Is the ML model making the decisions?
No. The model is advisory. Deterministic policies can block outright; risk scores above the threshold your team configures route decision-API transactions to human review, and on tool-call planes the score is recorded as an attributed detection. There is no path where the model alone denies or approves a transaction. The full posture, including verdict mapping and degradation behavior, is on the advisory risk scoring page.
On accuracy: the model is trained on public datasets, there is no public labeled dataset of agent-initiated transaction fraud yet, and we have no agent-traffic training data either. We therefore publish no detection-rate or accuracy claims for agent traffic. Every number we do publish carries the dataset and split it was measured on; the honestly labeled tables are on the scoring page. Thresholds will be calibrated with design partners against real agent traffic; until then, defaults are exactly that: defaults.
Does this replace our fraud or transaction monitoring system?
No. Your transaction monitoring, screening, and case management systems keep doing their jobs. This add-on governs the layer above them: whether an AI agent is allowed to initiate a transaction at all, under what mandate, and with what human oversight. It closes the gap those systems were never built to see: an authorized agent transacting produces none of the human behavioral signals those stacks read.
What data leaves our environment?
None through this add-on. Deterministic controls and risk scoring both run inside your deployment; the scoring service is a container in your environment, not an external API. There are no external data enrichment calls and no third-party data sharing, and the sanctioned-geography and merchant-category lists are versioned policy content applied by your team, not live lookups. If deployed on AxonFlow's managed service, this evaluation happens within your managed deployment under the platform's existing data handling terms. Raw prompt text is not sent to the scoring service at all; see data handling.
Which requests pause for approval, and which do not?
Pause-for-approval semantics apply on the decision API: step-ups produce a needs_approval verdict plus a pending approval queue entry. On tool-call planes, step-ups and above-threshold scores are recorded as attributed detections in the audit trail and nothing pauses, while deterministic block policies deny on every surface the pack evaluates on (the decision API and the MCP planes). If your integration needs the pause, call the decision API; the per-plane outcome table is the authoritative statement.
What happens when the scoring service is down?
The decision proceeds on deterministic controls alone, and the audit record states that scoring was unavailable (ml_inference_layer_status: "unavailable"). Degradation is visible, never silent: a reviewer can always distinguish a below-threshold score from an absent one. Deterministic blocks and step-ups are unaffected because they run in-process and fail closed.
Do we have to send transaction context for the add-on to do anything?
The transaction-aware controls read the documented transaction context objects, so they act only on requests that supply them, and the velocity and exposure step-ups additionally need the caller-supplied cohort aggregates. Two of the pack's controls bind to the request statement instead (the payment tool authorization gate and the payment execution step-up), so agent phrasings that touch payment routing or initiate payments are governed even without context. A context that is present but malformed does not pass silently on the surfaces the add-on covers: on the decision API it is escalated as unassessable, and on the MCP planes it is recorded as an attributed detection.
How do detections reach our auditors and regulators?
Every detection is a standard AxonFlow decision record with policy attribution, written to the same audit surface the platform's OJK, SEBI, and EU AI Act compliance exports read; see audit logging and evidence export. Human review entries carry oversight metadata, including EU AI Act Article 14 context and AML and CFT review context. Auditors see the control, the transaction, the outcome, and, where a step-up was reviewed, the reviewer.
Is this part of the source-available Community edition?
No. The AxonFlow platform is source-available under BSL 1.1, and the Community edition remains fully usable for building and testing governed agents. The Fraud & Risk Add-on is a commercial add-on to AxonFlow Enterprise, and it is not part of the base Enterprise license either; see Community vs Enterprise for the platform tier picture.
How is it priced?
It is a separately priced add-on on top of AxonFlow Enterprise. Charter partners receive early access and charter pricing. Contact us and we will walk you through it against your deployment size.
