Self-serve policy test

Can your AI agent spend $10 safely?

Set one owner policy, let the agent transact inside the limit, and block anything outside the limit automatically. This browser demo is a simulation, so no funds move.

Run the Demo See Pricing

$10Maximum single spend

$3 passesInside policy

$50 blocksOutside policy

agent-demo-01 Policy-controlled payment key

POLICY ACTIVE

Max spend$10

RailAuto

OwnerAudit

$3 payment

Agent simulates buying a small API call. It is inside the owner-defined policy.

$50 payment

Agent simulates a larger purchase. It exceeds the owner-defined policy.

Waiting for demo action Choose a simulated payment to test policy enforcement.

The audit trail will update below.

$3.00READYInside max_single_payment_usd policy.

$50.00READYWill be blocked above $10 owner limit.

--No payment attempt yet.

Real rail proof

Actual AgenticBTC transactions, not just a mock flow.

The browser demo above shows policy behavior without moving funds. These recordings show AgenticBTC using live payment rails for real transactions.

Strike live payment

AgenticBTC routes a real payment through the Strike rail with policy and audit context.

Coinbase USDC payment

Real USDC transaction proof for agent payments using the Coinbase rail.

NWC / Alby Hub payment

Non-custodial wallet connection proof for agent-driven Lightning payments.

What this proves

Policy once. Autonomous execution after that.

AgenticBTC is built for owners who want AI agents to transact without handing them unlimited spending power. The agent can execute automatically, but only inside the rules the owner configured.

{
  "agent_id": "agent-demo-01",
  "policy": {
    "max_single_payment_usd": 10,
    "allowed_rails": ["lightning", "strike", "coinbase", "nwc"],
    "manual_approval_required": false,
    "audit_log": true
  }
}

Build this into your agent.

Install the MCP package, connect your own rails, and give each agent a scoped payment key instead of a blank check.

Install MCP Package Read Setup Docs