How chat works
Start a thread. Write a message in the Space. Your teammates can reply, and your agents can be brought into the conversation with@.
Tag an agent. Type @ followed by an agent’s name (@Hops for the default agent, or @[name] for any custom agent) to bring them into the thread. They’ll respond from the context in this Space.
Tag multiple agents. A Space can have multiple agents. Tag different ones in the same conversation. Each responds from its own instructions, artifacts, and connected apps. Ask @Marketing for copy and @Legal for compliance review in the same thread.
Agents show their work. When an agent responds, it shows what it used and what it did. You can expand any response to see the full trace.
Threads stay together. Related replies stay in one thread. You don’t lose the context (who asked, who responded, what was decided) when you come back later.
Tagging agents
You can tag multiple agents in the same message or thread. Each agent responds only from the context and connections it has access to. If a tool isn’t connected to that agent, it will say so.
What agents can do in a thread
- Draft, summarize, and revise content
- Find and surface relevant docs or decisions from your connected tools
- Hand work off to another agent or teammate
- Flag actions that need your approval before they happen
Schedule from a chat
Any chat can have scheduled items. If an agent gives you something useful (a summary, a report, a check-in), you can schedule it to repeat on a recurring basis right from that conversation. This works in Space chats, agent chats, and even one-on-one conversations with the default Hops agent or your personal agent. See Scheduled for more on how scheduling works.Reactions and replies
- React to any message with an emoji
- Reply to a specific message to keep context tight
- Pin a message to make it permanently available to your agents. See Pins
Tips
Chat history is part of the Space’s shared context. Agents draw from it when you ask questions, not just when they’re directly tagged.