AVA Pay for WooCommerce
AI agents are already shopping your store. ChatGPT browses product pages, agentic checkouts are rolling out across the ecosystem. AVA Pay tells you which agents to trust, lets you set the rules, and records every verification and attributed order so you can see the traffic when reporting lands. This
AI agents are already shopping your store. ChatGPT browses product pages, agentic checkouts are rolling out across the ecosystem. AVA Pay tells you which agents to trust, lets you set the rules, and records every verification and attributed order so you can see the traffic when reporting lands.
This plugin connects your WooCommerce store to the AVA Pay verification API, which cryptographically verifies agent traffic across protocols (Visa Trusted Agent Protocol, IETF Web Bot Auth, Google AP2) through a single endpoint.
What it does
- Adds a verify endpoint ( /wp-json/ava-pay/v1/verify-agent ) that proxies signed agent requests to the AVA Pay API. Signatures are verified server-side against the agent platforms’ published keys.
- Applies YOUR policy: accept/reject verified agents, per-platform allow/challenge/block rules, discount caps, and spend limits, kept as a portable JSON policy document.
- Optionally mints a single-use, expiring WooCommerce coupon for verified agents.
- Shows signed AI agent visits to your store. When an AI agent loads a page with its requests signed (ChatGPT’s agent does; most crawlers do not), the plugin checks the signature with the AVA Pay API after the page has been sent and lists the result under WooCommerce, Agent visits: counts for the last 7 and 30 days by agent platform and outcome, and the last 50 visits. This only observes: a page view is never blocked, redirected or given a coupon because of it. You can turn it off in the settings.
- Records every verification and every attributed order in local database tables. (Agent page visits are shown as above; a traffic and revenue dashboard for the verify endpoint and orders is planned, and that data is not displayed yet.)
Trust model, honestly stated
- Verification proves agent identity and request integrity . A discount is only granted beyond your identity-only tier when the request carries a buyer mandate.
- Everything fails closed: if the verification API is unreachable, agents are not admitted (and the outcome is recorded as an error).
- We do not claim to have blocked an agent we never managed to check. When the verification API cannot reach an agent’s trust root, the request is still not admitted, but it is recorded as unverifiable rather than as a rejection, and the storefront response says verification_unavailable instead of agent_blocked .