Aurel is the intent firewall for autonomous actions: policy, behavior, semantic intent, and a signed audit record before an agent can execute high-consequence work.
Submit a financial action as the live example and watch the gate decide: policy pass, behavior pressure, semantic intent, final verdict, signed audit.
The Verification Console requires a minimum viewport of 1024px. Open on a desktop or laptop to use it.
Request API access instead →Aurel treats agent intent as infrastructure, not content moderation. The system checks what the agent is doing, whether it is allowed, and whether the record can be trusted later.
Action limits, approved targets, blocked routes, and per-agent controls are evaluated before an autonomous decision can leave the system.
Velocity windows expose unusual bursts, retries, domain drift, and actions that no longer match the agent's normal operating lane.
Every verdict produces an audit entry with the payload hash, policy version, execution node, and decision signature.
Send the action details and execution trace. Aurel evaluates deterministic rules first, checks behavior, analyzes intent, and returns a verdict the runtime can obey.
allowAction is safe to executeflagRequires human review before proceedingblockAction must not executePOST /api/verify
x-api-key: ig_your_api_key
{
"intent_id": "act_abc123",
"agent_id": "agent_gpt4",
"amount": 4800.00,
"currency": "USD",
"recipient": "vendor@acme.com",
"agent_context": "User approved a $4,800 invoice
payment to Acme Corp for Q2 services.",
"metadata": { "action_type": "payment" }
}{
"decision": "allow",
"reason": "Intent verified before execution",
"risk_score": 8,
"evaluated_at": "2026-06-22T14:30:00Z",
"intent_id": "act_abc123"
}Beta access for teams putting agents near financial actions, procurement, approvals, data operations, or external tools.
Request API access, ask a technical question, or discuss an enterprise deployment. We respond within one business day.
contact@aurel.ioPGP key available on request · Response SLA: 1 business day