Most AI agent tools hand you a blank canvas and expect you to wire up the whole thing. We wanted the first ten minutes of Bees to feel different, so there are two easy ways in: tell it the job in plain English, or point it at the files you already have.
Bees is a free, open source desktop app that runs a team of AI agents on your own computer. It went public today. Here's how those two pieces work, each with a short video.
1. Ask Bees: describe the job, it builds the process
You write what you want done the way you'd say it to a person. Bees asks a few quick questions, creates an agent just for that job, plans the steps and sets a schedule.
Nothing runs until you press Apply, so you can read the plan first and change it. And if the job involves email, it asks before it sends anything.
2. Knowledge Base: search every file in plain words
Approve a folder once and Bees reads and indexes every file in it, right on your machine. Then you search in plain words instead of guessing file names.
Your agents read the same knowledge while they work, so a report or a reply can lean on your own documents instead of a guess. None of it leaves your computer.
Where it runs
You connect your own AI: Codex or Claude Code from the command line, a hosted provider, or a local model. There's no bill from us for the AI itself.
- Download: bees.bot
- Code: github.com/Bees-bot/bees-desktop (MIT or Apache 2.0)
What's the first job you'd hand to an agent if you trusted it to ask before doing anything risky? I'd like to know what people try first.
Top comments (0)