Convoy AIAgents are already asking. Set the rules first.
ChatGPT, Claude, Gemini and other assistants now find, ask and act for your customers. A Convoy Station lets them in on your terms: the right data, the right steps, and a person or the customer where it counts.
Shoppers send their agent. You decide how it orders.
Agents ask for a product, a price and a delivery date, then try to order. Without governance, an agent orders twenty of a limited item, promises a delivery you cannot make, or places an order the customer never saw.
Your Convoy Station puts your catalogue, stock and delivery rules behind one question. It applies your limits per order, checks delivery before ordering, and lets the customer confirm before anything is placed.
Land on the shop, pick a category, stack four filters, open three products, check delivery, then start again. The catalogue is there. The answer is not.
- Reached the Terre & Table Convoy Station
- Called
search_products— olive oil, organic, ≤ 10 € - Checked live stock and delivery windows for Lyon 7 before ordering
- Added it to the selection for the customer to confirm
500 ml
One request. Real inventory, your rules, and the customer confirms before the order is placed.
An agent can book in seconds. Your intake rules still apply.
Agents ask to book, move and cancel appointments. Without governance, an agent takes the first free slot with the wrong practitioner, skips the questions you ask on the phone, or holds a large group booking with no deposit.
Your Convoy Station runs the booking as a playbook: qualify first, match against real availability, attach your cancellation policy, and ask for a deposit where your rules say so. Reminders and rescheduling stay in the same thread.
Pick a service, scan a grid of slots, open each practitioner to check the speciality, then start again when the slot is gone.
- Reached the Lumen Physio Convoy Station
- Asked the intake questions before searching
- Called
search_availability— evenings, this week, N1 - Matched practitioners by speciality, not by first free slot
- Checked insurance cover, then held the slot for the customer to confirm
Sports physio
One request. Real availability, the right practitioner, your policies applied before anything is booked.
Users send their agent into your app. It should see only what they may see.
Agents ask for reports, settings changes and refunds. Without governance, an agent reads data from teams its user does not manage, changes a plan without anyone noticing, or refunds whatever it is asked to.
Your Convoy Station exposes the queries and actions your product already supports, scoped to each person's role. Sign-in comes before private data, refunds above a threshold are approved first, and every call leaves a trace.
Read the grid week by week, switch to Timesheets, export four CSVs, pivot them in a spreadsheet, and hope you added it up right.
- Reached the Shiftly Convoy Station
- Checked the manager is signed in before reading timesheets
- Called
get_timesheets— team: Evening, 1–30 April - Compared scheduled against actual hours, week by week
- Scoped to the teams this manager is allowed to see, and traced
One question. Real timesheets, after sign-in, and only the teams this manager may see.