My AI agents have Trello accounts

My AI agents have Trello accounts.

Their own accounts. Their own names on the boards. Cards assigned to them like anyone else on a team.

This is how I solved the visibility problem. Agents work invisibly. I would ask for something on Monday and by Thursday have no idea whether it happened, stalled, or quietly died. Nothing to open, nothing to check. Just trust.

I manage agents the way I would manage people.

The tasks live on boards I can open any time. Who does what is written down in one GitHub repo of plain text files that every agent reads, whichever tool I open that day. Each morning, each agent goes through its due and overdue cards, works its own list, and only surfaces what needs a decision from me. The due dates decide what I hear about and what I never see.

I could buy an agent platform for this. Dashboards, orchestration, visibility included. The visibility is real and so is the lock-in. This setup is boring on purpose. If a better agent shows up next month, it gets a Trello account and reads the same repo. Nothing migrates. Nothing is held hostage.

Give your agents the tools you would give a person. A name, a task list, the rules in writing. Then watch them work.