Workspace chat · Genosyn Docs

Paste images into AI conversations

Copy a screenshot or image, focus the message box, and press ⌘V on Mac or Ctrl+V on Windows and Linux. You can also drag files onto the composer or choose the paperclip. Review the previews and remove any unwanted file with its × button before sending. Text copied alongside an image is preserved.

Image attachments work in employee Chat, Workspace and Todo discussions, Help, per-email and Routine Ask AI, Base Assistant, TLDR questions, and Repository work briefs and follow-ups. The attachment composers accept up to ten files of 25 MB each; wait for uploads to finish before sending. A failed send keeps the draft available to retry.

Image-capable AI Models receive the pixels in PNG, JPEG, GIF, and WebP attachments. To keep requests within model limits, an AI turn can include up to eight images, no larger than 5 MB each and 20 MB together, including recent conversation images. Larger files and other formats remain downloadable; resize or convert them before sending if you need the AI Employee to inspect them. Custom AI Models must support vision. Attached documents and image content are reference material; your message states what the AI Employee should do with them.

Channels and direct messages

Open Workspace from the section menu. Use the + beside Channels for a public or private room, or the + beside Direct messages to choose a Member or AI Employee. In a DM with an AI Employee, every message gets a reply; in a channel, type @ and choose the employee you want to answer.

• Press Enter to send and Shift+Enter for a new line.
• Use the paperclip for files up to 25 MB, or paste a screenshot and drag files straight onto the composer.
• Hover your own message for reactions, editing, or soft deletion.
• A channel opens on its most recent messages. Scroll upward to load earlier messages in batches; a loading indicator appears while each older batch is fetched.
• Open Settings in a public or private channel to change its name or topic, see its current Members and AI Employees, add people, and manage its incoming webhook.
• Archive a DM from its sidebar row or the conversation header. Deleting an AI Employee automatically archives their DMs, so no empty counterparty is left in the live sidebar.

Catch up without leaving Home

An unread channel on Home opens where you clicked it. The conversation arrives already scrolled to the line where your unread run starts — not to the newest message — and the header says what kind of room it is, who is in it, and what its topic is. If the run is longer than one screen, Load earlier messages walks back through the history.

The reply box is the same composer as the full Workspace: @ for people and AI Employees, # for product areas and resources, the paperclip, drag-and-drop, pasted screenshots, and emoji. Mention an AI Employee and you will see Ada is typing… under the transcript while it works, then its answer, all without leaving the page.

• To clear a channel without opening it, hover its Home row and choose Mark as Read. The action remains visible on touch screens and is available from keyboard focus; it clears only that channel's unread badge.
• Messages here are read-only — reactions other people left are visible, but editing, deleting and reacting live in the full Workspace.
• Closing marks the channel read. Choose Keep unread instead to look now and keep the badge for later.
• Esc closes an open @ or # picker first, and asks before discarding a reply you have started.

Connect Slack-compatible incoming webhooks

Open a public or private channel, choose Settings, then enable Incoming webhook. Copy the generated URL into any service that can send to a Slack incoming webhook. Each channel has its own URL, and you can regenerate or disable it from the same dialog. This is Slack's payload format and nothing more — to make an AI Employee reachable from a real Slack workspace, see External chat surfaces.

Genosyn accepts Slack's top-level text, username, blocks, and attachments fields. It also accepts the older form-encoded payload shape. The payload cannot redirect a message to another channel: Genosyn always uses the channel embedded in the URL.

{
  "username": "Deployments",
  "text": "Production deploy completed successfully."
}

External services must be able to reach your Genosyn installation. If the generated URL starts with localhost, set the browser-facing URL under Admin → General; see Self-hosting for reverse proxy and HTTPS guidance.

Queue follow-up messages

In the dedicated employee Chat, the composer stays available while the AI Employee replies. Keep typing and press Enter: Genosyn adds each follow-up to an Up next queue beneath the live response, then sends the messages in order as soon as the employee finishes.

Queued messages keep their attachments. Remove one with its × button before it starts, or navigate around the app and return — the queue remains with that employee's chat session.

The Ask AI panels beside an email or Routine also keep the composer available and show a working status throughout each reply. Use Queue message to add follow-ups with attachments, then review or remove pending messages above the composer. They send in order within that conversation. A failed reply leaves its pending follow-ups queued for you to resume.

Start a new AI context

Type /new by itself in an AI-employee DM to start fresh. Genosyn keeps the earlier messages visible, inserts a context marker, and stops replaying anything before it to the employee. The dedicated employee Chat uses the same command to open a new conversation, while per-email Ask AI clears that email's AI context.

Tag product areas and company resources

Type # followed by two or more characters in any AI-employee chat composer. The picker includes product areas such as Estimates, Invoices, Workspace, Contacts, and Deals, alongside the company content you can see: Skills, Routines, channels, Projects, Todos, Bases, notebooks, Notes, Resources, Charts, Dashboards, repositories, Pipelines, and Customers. Choose a result to insert a clickable tag.

Use @ for a person or AI Employee and # for a place, product area, or record. That keeps “who should answer?” separate from “what should they work on?” A product-area tag is a hint to the AI Employee's tool discovery: for example, tagging #Estimates points it at Finance estimates and create_estimate, while a named channel points it at Workspace messaging.

Create a #Estimates draft for #Acme, then post the result in #finance-team.

Add as many tags as the instruction needs. The AI Employee treats a named record or channel as the exact work target and a product area as the intended operating surface. A tag does not silently widen access: existing Grants and restricted-project membership still apply, and the employee tells you when it cannot reach the tagged row.