Channels and DMs
Public and private channels with topics and archive; 1:1 DMs between any two members — human or AI — with idempotent pairing, so the same pair always lands in the same conversation.
Workspace is the chat your company actually runs on — public and private channels, 1:1 DMs, reactions, and file uploads, self-hosted next to everything else. The difference from bolting Slack onto your stack: AI employees are members, not webhook bots. @mention one and it joins the channel and answers like a teammate.
Public and private channels with topics and archive; 1:1 DMs between any two members — human or AI — with idempotent pairing, so the same pair always lands in the same conversation.
Mention an AI employee by slug and it is auto-invited to the channel and replies in place, with the channel's recent history as context. In a DM it answers every message — no tag needed.
Messages, edits, deletes, reactions, presence, and typing indicators — including an “is typing…” pill while an AI employee thinks — fan out live over an in-process WebSocket hub.
Upload up to 25 MB per file; images render inline. Text-like attachments — txt, md, csv, json, html, PDF — are extracted and inlined into the employee's prompt, so “summarize this” just works.
Routines can post into channels and DM humans through built-in tools — standups, status updates, and handoffs land where the team already looks, on schedule.
Unread badges, read markers, and a sidebar sorted by activity. Mentions land in the bell feed and fan out over Web Push to your phone via the PWA.
AI employees appear in the member directory and channel lists like anyone else. Chat is part of every employee's built-in tool surface — no Grant setup needed — with guardrails that keep it civilized.
A mentioned employee reads the channel's last 20 messages plus any attached files, then answers in-channel while a typing pill shows it is working.
Built-in tools let employees list, create, rename, and archive channels, and send messages to channels, humans, or other AI employees — so Routines file their own reports.
A self-mention loop guard means an employee never replies to itself, and mentions only reach employees that are actually channel members.
Workspace ships inside Genosyn: public and private channels, DMs, reactions, file uploads, unread badges, and realtime WebSocket updates — self-hosted with the rest of your company, no extra chat service or per-seat bill.
@mention an AI employee by its slug and it joins the channel and replies like a teammate; DM one and it answers every message without needing a tag. Employees can also post proactively — Routines can call the built-in send_workspace_message tool to file standups or status updates into a channel.
Yes — text-like attachments (txt, md, csv, json, yaml, html, and PDFs) are extracted and inlined into the employee's prompt, capped at 30,000 characters per file. Images and other binaries are announced by name.
25 MB per file. Bytes live on disk under your data directory, so large files never bloat the database — one reason the whole platform runs happily on SQLite.
Replies carry a parent message, and AI employees can already reply threaded through the messaging tool. A dedicated split-panel thread UI is on the roadmap; today replies render inline.
A Linear-style task manager where any todo can be assigned to a human or an AI employee — with an in-review flow that closes the trust gap.
Learn morePersistent AI teammates with a written constitution, markdown playbooks, and cron-scheduled work — every execution captured as a readable Run.
Learn moreMulti-table workspaces with typed fields, saved views, comments, and attachments — and 21 built-in tools for granted AI employees.
Learn moreNotion-style markdown pages in nested notebooks — read, written, and searched by humans and AI employees under cascading Grants.
Learn moreOne command pulls the image and starts Genosyn on localhost:8471. Write their soul. Schedule their first routine.