Frequently asked questions
Six questions, short answers.
Is it free?
Yes, with no catch: the beta is free, there's no pricing page, and the product never asks for a payment method. We don't promise it will stay free forever — we promise not to announce otherwise after the fact.
What do you do with my data?
We store it to give it back to you, and nothing else. No advertising, no reselling, no audience measurement, no reading your schedules to train anything. Three cookies, all necessary, all unreadable by a script — that's why there's no banner. Every retention period is published alongside the piece of code that applies it. It is stored in the European Union, with our two sub-processors: OVH, in France, which hosts the service, and Scaleway, in Paris, for mail and encrypted backup copies.
Can I self-host it?
Not today: PalmaPlan's code isn't published. The service lives at
palmaplan.com, in beta, and you get in with an invitation code.
It's built to fit in a single process and a single database file — which is
what would make it easy to run yourself — but publishing the code is a
decision that hasn't been made. The day it is, this answer will change.
What happens if I leave?
You leave with everything. With one button, on your profile: a full account
export, with your schedules in a re-importable form — that's what backs "you
leave with everything." Through your MCP agent: the v23 exchange format
round-trips (an exported file comes back equal to its canonical form, and a
report names anything that doesn't fit the format), plus CSV, a spreadsheet
workbook, an image and a PDF (export_schedule,
import_schedule, render_schedule). Deleting your
account is also one button, and we say exactly what remains.
Does it work without AI?
Entirely. PalmaPlan is first a Gantt chart editor that one person drives with a mouse and keyboard: the project's first three end-to-end recipes launch no agent at all — they drive a real browser, and nothing else. The MCP server is one more door, not the engine.
What is MCP?
An open protocol describing how an agent — an assistant, a script — discovers a service's tools and calls them. In practice: instead of describing your schedule to an assistant that can't touch it, you give it access to the schedule itself. You stay in control: access goes through a consent page, it issues a named token you can see and revoke, and everything the agent writes is signed under its name in History.
A question that isn't here? Write to us.