Coworkers

Create focused chat personas — each with its own memory, instructions, and sharing — in the Logic AI sidebar.

Coworkers are focused chat personas you create in the Logic AI sidebar. Each coworker is a named assistant with its own memory, its own colour, optional standing instructions, and its own set of chats — so you can keep a “Renewals” coworker separate from a “Data cleanup” coworker instead of running everything through one general chat.

A coworker isn’t a different AI — it’s the same Logic AI, scoped to a job. It runs in your user context with all the same permissions, tools, and approvals as any other chat.

Why use coworkers

• Separate contexts — keep unrelated work in its own lane. A coworker only ever sees the chats and memory that belong to it.
• Standing instructions — give a coworker a persistent role once (“You help me manage EMEA renewals; always check the contract end date first”) instead of repeating it every conversation.
• Its own memory — each coworker remembers facts relevant to its job, and nothing bleeds between coworkers or into your general chat.
• Shareable — make a coworker available to specific teammates or your whole org, so everyone works from the same persona and shared memory.

Creating a coworker

In the chat sidebar, click + New coworker and fill in:

• Name — what it does, e.g. Renewals, Support triage, Data cleanup (required, up to 80 characters).
• Instructions (optional) — standing guidance applied to every chat with this coworker. Think of it as a mini system prompt for this persona.
• Colour — a swatch so you can spot the coworker at a glance. Leave it and Logic AI auto-assigns a distinct colour from the palette.
• Visibility — who can use it (see below).

Your coworkers appear as chips in the sidebar. Click one to start (or continue) a chat with it; its colour tags the chats that belong to it.

Visibility & sharing

Every coworker has one of three visibility settings:

Visibility	Who can use it
Private	Only you (the owner). This is the default.
Specific users	You, plus the teammates you pick.
Org	Everyone in your org who has Logic AI.

Notes on how sharing works:

• When you choose Specific users, a search box lets you add teammates. Only users who actually have Logic AI assigned (one of the Logic AI permission sets) appear in the search.
• Only the owner can rename, recolour, re-share, or delete a coworker.
• A shared coworker shares its memory with everyone who can use it — that’s the point, so the team works from the same context. But chats are private to each user: even on a shared coworker, you only see your own conversations, and the “3 chats” count next to a coworker reflects your chats with it.

Coworker memory

Each coworker keeps its own memory, separate from your personal memory and from other coworkers. It’s captured the same way the rest of Logic AI’s memory works — automatically, as you chat — and recalled invisibly in future conversations with that coworker. See Memory for how capture and recall work.

You can review a coworker’s memory from its menu in the sidebar:

• View the facts it has learned.
• Edit a fact to correct it.
• Forget a fact you don’t want it to keep (it’s tombstoned so the capture layer won’t silently re-learn it).

Because a shared coworker’s memory is shared, anyone who can use the coworker can curate its memory — a small, collaborative shared brain for the team.

Owners can also regenerate a coworker’s guidelines — a short, model-written summary of how the coworker should work, distilled from its memory. Guidelines refresh automatically after memory consolidation; the owner can trigger a refresh manually from the coworker’s menu.

Deleting a coworker

Deleting a coworker (owner only) removes the coworker and purges its memory. Your existing chats aren’t deleted — they’re simply unlinked from the coworker and fall back into your general recent chats list.