for AI agents.
Give an agent a wallet it can't overspend. Per-call payments, hard on-chain rules, instant settlement.
Set a policy - daily $50.00, per-call $0.05, an allowlist of services - and a smart contract enforces it on every transaction. The agent literally cannot exceed its budget.
No bank gives software an account with cryptographic spending limits.
A smart contract does. Each guarantee below isn't a promise - it's enforced mechanically. Watch the rail work.
Programmable rules
Spending limits live in a smart contract, not a dashboard setting. A payment over the per-call cap doesn't get flagged - it bounces at the gate and never settles.
Micropayments to anyone
Agents pay a fraction of a cent per call in stablecoin, settling instantly and fanning out worldwide - exactly where ordinary rails break.
Auditable spend
Every payment is signed, on-chain, and traceable to the policy that allowed it. The receipt and the rule that approved it are one source of truth.
- 0.02 USDCservice://api.search0x9b1f…
- 0.02 USDCservice://api.embed0x9b1f…
- 0.02 USDCservice://api.rerank0x9b1f…
- 0.02 USDCservice://api.vision0x9b1f…
One rail, two perspectives.
one rail - the builder sets the policy, the agent spends within it
For builders
You ship the agent. Railo gives it a wallet with rules you control and a clean way to pay for the services it depends on.
- Drop a wallet into your agent with one SDK call.
- Express spend as policy - limits, caps, allowlists.
- Meter any API behind a service:// address.
- Reconcile from on-chain receipts, not invoices.
Give your agent a wallet it can't overspend.
Spin up an agent, set its policy, and make a metered payment - the whole loop lives in the developer guide.