Message it like a colleague
Write to a bot the way you write to a teammate. It works the task through, shows every step, and comes back when it is done.
New guideBotseon 101
Bots that do the job, on a computer of their own.
Give a bot a job description, sign it in to your tools, and check its work when it asks.
The app: a sidebar of five bots and the Invoice Chaser transcript ending in an approval card asking whether to send a reminder about invoice 2041.
See pricing: €0/monthFree€30/monthPro€100/monthPro+
What a bot does
What a bot does, shown in the product itself.
It asks before it acts
Sending, paying, deleting: a bot stops and asks, with the reason on the card. Say once, always, or no. An automatic review checks every action as well.
Teach it a task once
Do a job once while the bot watches. It keeps the steps as a skill and runs them on a schedule as a routine.
It remembers
What a bot learns is stored as facts you can read, edit and delete, never as instructions it must obey.
Not tied to one model
Any model can power a bot. The managed service runs the models on EU routes and meters every call at a published price, in credits you can count. The framework runs on your own keys or a local model.
It works on a computer, not on yours
Your bots share one computer per person, each with a screen of its own: a browser, a terminal and files. A bot signs in to your tools and uses them the way you would, including the ones with no API.
How it works
From sign-in to a bot at work, in three steps.
Sign in and name a bot
A sign-in link by email. Your own organisation is created with it.
Give it a job
A job description in plain words, and the tools it may use.
Check its work
Every step in the transcript; every question on a card you answer.
Give each bot a job
Five starter bots ship today. Each one is a job description you can change.
Where your data lives
EU/EEA storage, no non-EEA egress by default, US sub-processors under SCCs.
- Frankfurt
- Database and files: Supabase, eu-central-1
- Frankfurt
- Web tier: Vercel, fra1
- Frankfurt and Amsterdam
- Bot computers: Fly.io, fra and ams
You are the controller for the people your bots read, remember or contact. Botseon is the processor for that content, and the controller for your account, billing and security data.
Sub-processorsBotseon today
- 187
- documentation pages published
- 6
- connectors ready to use
- 5
- starter bots on day one
Questions
Short answers before you make a bot.
How is a bot different from an assistant?
A bot works on a computer in the cloud, signs in to your tools, and keeps working while you are away. It asks before it acts, and you can read everything it did.
Who is it for?
Small teams that want a job done without building a workflow first, and developers who want to run the framework themselves.
Where is my data stored?
EU/EEA storage, no non-EEA egress by default, US sub-processors under SCCs. The database is in Frankfurt; the sub-processor list is on the legal page.
Which models does it use?
Any model can power a bot. The managed service runs models on EU routes and shows you the price of each call; the framework runs on your own keys or a local model.
What does it cost?
A monthly plan per person or per seat with a weekly allowance of credits, and a hard cap on by default. The prices are on the pricing page.
Do bots share one computer?
One computer per person by default, shared by that person’s bots.
Can I run it myself?
The framework is open source under Apache-2.0 and free to self-host. Its public repository arrives in a later release.
Your bots, one click away
The desktop app opens on the same account as the browser, so every bot and every conversation is already there.
Desktop
Builds are being published, and this is the page they will appear on.
Phone
- iPhone · Not published yet
- Android · Not published yet