Veyto
Give an AI agent a wallet with enforceable spending rules, and get a signed receipt for every payment it makes.
Wallets · Guardrails · Receipts.
Veyto sits between your agent and the thing it wants to buy. Your agent asks to pay; Veyto checks the limits you set; if the payment is allowed, your own wallet signs it and the money moves. If it is not allowed, nothing moves and you get told why.
You keep your keys. Veyto never holds them, and could not spend for you if it wanted to.
Start here
Quickstart — from an API key to a real payment on Base Sepolia, in seven steps. It is the fastest way to see the whole thing work.
Then run the complete example, which does all of it in one file you can copy.
The rest of the documentation
| Concepts | The five words Veyto uses, and what each one means |
| Authentication | Your API key, and what it can do |
| Agents | Creating an agent and pointing it at your wallet |
| Guardrails | The three limits, and freezing |
| Governed payments | What happens during a payment, step by step |
| Blocked | What a refusal looks like — and why it is a success |
| Receipts | What Veyto signs, and what each field means |
| Verification | Checking a receipt without trusting us |
| Errors | Every error the API returns |
| Limitations | What the beta does not do |
Two things worth knowing early
Blocked is not an error. When a guardrail stops a payment, that is the product working. Veyto reports it in blue, not red, and no money moves. See Blocked.
A receipt proves itself. Every payment produces a signed record that anyone can check at verify.veyto.io — no account, no key, no permission from us. That is deliberate: a receipt you can only verify by asking us is not evidence.
This is a beta
Base Sepolia test network and test USDC only. No mainnet, no real money. Read Limitations before you plan around it.