Accept AI agents without the risk. Every purchase is checked against limits the human signed, decided by exact rules, and reconciled against what was actually charged. Wardpact never touches the money.
Today a merchant that sees an AI agent has two options: trust it blindly or block it. Wardpact adds a third — let it buy, only within what the human signed, and prove it.
A person approves a mandate on their phone: how much, where and until when. The agent never holds passwords or card data — it carries the signed pact.
Each purchase is checked against the pact with exact, deterministic rules — no LLM guessing — and approved purchases get a signed proof any payment rail can verify.
The payment goes straight from the customer to the merchant. Wardpact compares what was authorized with what was actually charged and flags any mismatch.
Inside the merchant's app or the assistant itself, before it can spend a cent.
Plain limits: how much, where, until when — and which purchases need their OK first.
Face ID, fingerprint or device PIN. The signing key stays on their phone — Wardpact, the agent and the merchant never see it.
Alerts on suspicious attempts, a tap to confirm bigger purchases, and one place to see and revoke every pact.
The same request always gets the same decision, with a reproducible record of why.
Many agents buying at once never push a pact past its limits, in any currency.
Authorized versus charged, checked automatically. Over-charges are flagged, cancellations release the budget.
One policy and one audit trail across agents, models and payment protocols — not tied to a single network.
Tamper-evident, signed records your compliance team can verify — without storing personal data.
Wardpact never processes payments, holds funds or touches card data. It authorizes and proves.
We're working with a small group of fintechs, merchants and platforms to shape Wardpact. Leave your email and we'll reach out for a 20-minute conversation.