ALLOW
Low-risk, in-policy actions proceed without added friction when signals and limits are satisfied.
GuardRoute on one payment
One fact. The payment stops.
Same $4,200. Change the payee and ALLOW becomes ESCALATE.
Known payee
Northline Custodial
$4,200
ALLOW
New payee
Harbor Relay LLC
$4,200
ESCALATE
Live product
Live product · The control layer for high-risk actions before execution
ProofAudit proves what happened. GuardRoute controls what happens before it happens.
Public demo · illustrative data
Policy decision panel
ActionOutbound wire
Amount$240,000
New beneficiaryYes
GeographyUnusual for this account
Velocity3× normal
Risk score82 / 100
Policy resultHOLD
Required actionTreasury + Compliance approval
Reason codes3 — BENE_NEW, GEO_ANOMALY, VELOCITY_3X
ReviewerTreasury lead (pending)
Final outcomeHOLD · REQUIRE APPROVAL before execution
Control outcomes
ALLOW WITH LIMITS
Approved within caps — amount ceilings, time windows, or scoped permissions that policy defines.
REQUIRE APPROVAL
Named approval is required before execution when amount, beneficiary change, or sensitivity crosses policy.
ESCALATE
Named owners must review before execution when score or rules exceed standard thresholds.
HOLD
Execution paused pending investigation, callback, or document verification — not a silent queue.
BLOCK
Hard stop when policy or fail-closed rules require no release until risk is resolved or overridden with record.
Why teams buy this
- Stop a high-risk action before it executes
- Use the same policy for people and AI agents
- Let in-policy actions proceed
- Name the reviewer when approval is required
- Keep the reason for the decision on the record
Control boundaries
- GuardRoute decides the control. It does not hold funds or execute the action.
- It does not replace fraud, compliance, or treasury judgment.
- ProofAudit proves what happened. GuardRoute controls what happens before it happens.
Explainability
Every routing decision is designed to retain enough context for treasury, fraud, compliance, and audit partners to understand why a path was chosen — without reconstructing chat or email threads.
Decision record (schema)
- risk_score
- Numeric score and contributing signal summary
- triggered_rules
- Policy rules that fired for this action
- policy_reason
- Human-readable rationale tied to rule IDs
- reviewer
- Assigned or completed reviewer identity
- override_history
- Overrides with reviewer, reason, and timestamp
- final_outcome
- ALLOW, ALLOW WITH LIMITS, REQUIRE APPROVAL, ESCALATE, HOLD, or BLOCK
AI-agent controls
When an AI agent attempts a sensitive action — moving money, changing a beneficiary, releasing information, or triggering another high-risk workflow — GuardRoute determines which controls must apply before execution.
Agent requests outbound wire to a new beneficiary
HOLD · REQUIRE APPROVAL — execution waits for Treasury and Compliance
Agent updates vendor ACH instructions before a scheduled payment run
ESCALATE — beneficiary change; callback verification required
Agent exports a batch of customer account records to an external workflow
BLOCK — data-release policy; no release until security review
Agent initiates a recurring payment under an established allowlist
ALLOW WITH LIMITS — inside the velocity and amount envelope; control decision record written
Control architecture
When an AI agent attempts a sensitive action, GuardRoute determines which controls must apply before execution. The evidence record is the control decision. The document package, when the products run together, is ProofAudit.
Action Request
Risk Signals
Policy Evaluation
Decision
Human / Automated Control
Execution
Evidence Record
Use cases
- Payment approvals
- Beneficiary changes
- High-risk workflow actions
- AI-agent permissions
- Compliance escalation
- Fraud review
- Exception management
- Transaction controls
Integration patterns
Specific connectors are confirmed during scoping.
Payment and banking systems
Wire, ACH, and treasury initiation points where a control gate must sit before release.
Core / LOS / workflow systems
Loan, account, and operational workflows that trigger sensitive state changes.
Identity and approval tools
Approver directories, step-up authentication, and entitlement checks in the routing path.
Case management
Fraud, compliance, and exception queues that receive escalated route decisions.
APIs and webhooks
Programmatic decision requests and asynchronous disposition callbacks into your stack.
FAQ
- Is GuardRoute live?
- Yes. GuardRoute is a live product. Book a demo for a guided walkthrough. The public demo uses illustrative data, not customer data.
- Does GuardRoute execute payments?
- No. GuardRoute decides the route before execution. Your payment, banking, or workflow systems execute once policy and controls are satisfied.
- Can humans override a route decision?
- Yes. Authorized reviewers can override when policy allows. Overrides are recorded with reviewer identity and reason.
- Does GuardRoute replace our fraud or compliance team?
- No. GuardRoute routes work to the right desk — fraud, compliance, treasury, or standard path — with a clear disposition before anything runs.
- How does GuardRoute work with AI agents?
- When an agent requests money movement, beneficiary changes, data release, or another sensitive workflow step, GuardRoute scores the action and returns required controls before your systems execute.
- How is GuardRoute different from ProofAudit?
- ProofAudit proves what happened. GuardRoute controls what happens before it happens.
Built by StratEdge Workflow Systems
Enterprise workflow and AI control infrastructure for regulated and document-heavy financial operations.
stratedgeworkflow.com →Sister products