AI workspace
An AI workspace is a persistent, shared place where a team does its AI work. Instead of each person’s chats living in a private sidebar, the conversations, files, background material and agents for a project or team sit together, and everyone with access can see and build on them.
Why it matters
A chat window is built for one person and one question. Real work runs over weeks and involves several people, and the useful parts of it (the brief, the decisions, the files, what the agent already tried) need to outlast any single conversation. When they don’t, every session starts with someone re-explaining the background, and a new teammate has nothing to pick up.
A workspace also gives access rules a natural home. It’s easier to say “this project’s files are visible to these people and their agents” than to manage permissions chat by chat.
What a good one has
- Persistent context that is loaded once and stays current, not pasted into each conversation.
- Clear membership, so it’s obvious who, and which agents, can see what.
- Work you can track: threads for discussion, files for output, and a shared to-do list so tasks don’t fall through the gaps between people.
The risk is the opposite of single-player chat: a workspace that collects everything becomes a junk drawer. Keep one per project or team, and archive it when the work ends.
How elvex relates
In elvex, work lives in Spaces, which are projects or teams. Threads, files and to-dos live inside them, and the agents people connect work there alongside the people.