“Create a customer brief.”
Point your agent at a customer record in HubSpot. It reads the account, cross-checks the facts, and returns a deck you can read in five minutes.
→A cited research brief you review before it goes further.
Redeep is a native macOS workspace for handing research, operations, and coding tasks to AI coworkers. Track them on a board and review every result before it ships.
Turn a pile of tasks into visible progress: set up a board, add the work, let your AI coworkers take it forward, then review what comes back.
Your AI work, visible at a glance.
Each card is an isolated session, so parallel work stays organized instead of disappearing across chat tabs. How boards work →
Create a board for your project and assign a manager you can message on Slack. Your team has a home base.
Add a task in plain language and assign it to a coworker. Every piece of work has a place and a next step.
Multiple coworkers research, code, and use tools in parallel. Watch tasks move across the board while you focus elsewhere.
Behind the scenes: each task runs in its own isolated session.Open the completed card, inspect the report, diff, or dashboard, and decide what happens next.
Redeep isn’t just for code. Hand off the kinds of work you’d give a smart teammate — research, analysis, writing, ops — and come back to a finished deliverable you can review.
Point your agent at a customer record in HubSpot. It reads the account, cross-checks the facts, and returns a deck you can read in five minutes.
→A cited research brief you review before it goes further.
Drop in a messy spreadsheet. Your agent cleans it, runs the numbers, and comes back with the trends that matter and the chart that explains them.
→A clean dataset, a chart, and the findings — ready to check.
Give it the talking points and your voice. You get back a launch announcement, ad variations, and matching social posts you can edit or ship.
→A campaign’s worth of copy in your voice — review and ship.
Hand over a raw list and the criteria. Your agent enriches each account, scores them, and drafts a tailored first-touch email for every one.
→A scored, enriched list with a draft per account — you approve.
Hand it a bug or a complex feature. It works in an isolated branch, runs the tests, and returns a diff for you to review and merge. For complex requests, the agent will create cards it can follow up on.
→A working branch with a diff you review before it lands.
Ask it to publish what it just built. Your agent packages the work and serves it from your Mac with a secure link — no hosting bill, no deploy pipeline.
→A live link your team can open — you decide who sees it.
Bring your own API keys. Route across five cloud providers and on-device MLX models (open-weight or frontier) with no markup.
How model keys workIt browses the web (with an embedded browser) and drives any Mac app via computer use — using your existing logins.
How computer use worksManage your team on a Kanban board where each card is its own isolated session. Queue, parallelize, and automate work — or hand a card to @codex.
How boards workPackage know-how as reusable skills, fan work out to sub-agents, and connect your stack over MCP — with 26 starter skills included.
How skills workEvery capability below is built in and runs on your machine. Follow a link for the full technical breakdown.
Redeep lives & works on your device. Code and credentials stay on your machine. Read the full security breakdown →
Connect your favorite model provider to get started.
Connect a Google account in Settings → Connections and your AI coworkers can work with the mail and calendar a task actually needs. Redeep uses only the permissions you approve at sign-in, and anything that changes your data waits for your approval first. The same connection covers Docs, Sheets, and the Drive files Redeep creates or opens. Disconnect at any time to end access. See exactly what each permission does → Read the privacy policy →
Download Redeep, connect a key you already have, and let your team of AI coworkers go deep on your work. Free.