Convoy AIOne mission.
The interface layer moves to the agent. Every product and every team will need a control station for the agents that reach it — so we build Convoy Stations.
What we are building.
Convoy Stations are control stations for AI agents: agents plug in, rules apply. A Product Station lets customers send their own agents — ChatGPT, Claude, Gemini, Muse — to a store, a SaaS or a booking service, under the vendor's rules. A Process Station lets employees' agents reach the CRM, the ERP, databases, Slack and Jira, scoped to their role, with approvals, data kept inside and every call traced.
We started it because the gap was obvious and nobody was closing it. Systems either give agents nothing to call, or open far too much to them. Don't open your systems to agents until you set up a Control Station.
How we work.
Small, and close to the people using it. You connect your APIs, we build your station — skills, playbooks, policies — and we expose it on AI channels with you, watching the first sessions together rather than sending a form back.
We are based in France and build for European rules from the start: consent before measurement, decisions logged and explainable, data processed and stored in Europe on demand.
Product and platform. Works on what a station exposes, how it is governed, and what customers and teams are actually able to delegate to their agents.
LinkedIn ↗sokhna@convoy-ai.comEngineering and integrations. Works on the station runtime, the connectors, and keeping up with standards while they are still moving.
LinkedIn ↗mame@convoy-ai.comNotice and privacy
Who publishes this site and what we do with your data is set out on the legal page.