One workspace for the whole job — not five tabs glued together.
Tasks, chat, files, docs, and time tracking in one place, built on a fast Postgres core. Then an AI layer on top that actually does work — on your key, governed by you.
The PM core — solid, fast, and complete.
The fundamentals done right, so the AI has something real to stand on. All of this ships and works today.
Tasks & projects
Structured tasks with assignees, due dates, priorities, tags, and subtasks. Group work into projects; view it as a list, a board, or a timeline.
- · Board · list · timeline views
- · Subtasks, dependencies, recurring
- · Per-project keys (WEB-128)
Workflows & approvals
Move work through stages your team defines. Gate sensitive steps behind an approval so nothing ships without the right sign-off.
- · Custom statuses per project
- · Approval gates with an audit trail
- · Automations on status change
Team chat & DMs
Channels per team and project, plus direct messages — all with @mentions, files, and voice notes. Conversation lives next to the work it is about.
- · Team · project · custom channels
- · Direct messages + group threads
- · @mentions that notify
Files & folders
Attach files to tasks, chats, and projects, or keep a shared folder tree. Voice notes transcribe inline. Signed-URL storage — no public buckets.
- · Drag-drop on any surface
- · Nested folders, vers-safe
- · Inline voice-note playback
Wiki & docs
A nested doc space per team for specs, runbooks, and decisions. Auto-save, rich text, and provenance (who created, who last edited).
- · Folders + nested pages
- · Auto-save rich text
- · Created-by / edited-by provenance
Admin & governance
Custom roles, granular permissions, multi-team workspaces, and an immutable audit log. Built on Postgres row-level security, not bolted-on checks.
- · Custom roles + permissions
- · Multi-team membership
- · Append-only audit log
AI that does the work, governed by you.
Three surfaces, one principle: AI proposes, you stay in control. Bring your own key with zero markup, or let us manage it.
AI co-pilots
Smart Cmd+K turns plain English into a structured task. Clarifying questions when a request is ambiguous. Summaries, smart replies, and a personal daily digest scoped to what you can see.
Governed agents
Agents are real team members with a role, a token budget, and a trigger set — but they propose before they act. The Trust Ladder decides what runs automatically and what waits for your approval.
MCP — your tools, our data
A full Model Context Protocol surface at POST /mcp. Bring Claude Code, Cursor, Cline, or your own client — read tasks, post comments, change status, start timers, run agents. Same auth, same data.
An agent is a teammate with a budget — and a leash.
Every agent is a real team member with a role, a monthly token budget, and a set of triggers. It can be scoped to one project, held to propose-only, or trusted to act — and every action it takes is in the audit log. You decide how much rope, one rung at a time.
Reviewer Bot
✨ AI agentWatches PR-linked tasks · posts review comments
- @mention In any comment on watched tasks
- webhook GitHub PR opened / updated
- cron Off · enable to send daily summaries
auth/refresh.ts.
Want the full AI architecture?
BYOK vs Managed, the MCP surface, the no-training promise.
Ship the next thing faster.
Free for ten users. No card required. AI optional.