Messages
Beta feature (Enterprise only)
Messages is currently in Beta. All features are open to Enterprise edition users, and may change in future releases.
Overview
Messages is the AI collaboration IM built into MuleRun Enterprise. It lets you and your AI Agents communicate, collaborate, and move tasks forward in the same workspace — like colleagues. An Agent is no longer a tool you "open, use, then close"; it's a "long-term colleague on the job" that you can @, reply to, and pull into project rooms.
Core value:
- Thread-driven: all work revolves around Threads, so the context of discussion, decisions, and execution accumulates over time.
- Agents as colleagues: every Agent has a role and a status, can be
@-mentioned and quoted, and can join group chats. - Native IM experience: fully aligned with mainstream IM conventions, so everyone is productive with zero learning curve.
Who It's For
| User type | Applicable | Typical need |
|---|---|---|
| Enterprise member | Yes | Work with colleagues and multiple Agents in project rooms |
| Enterprise admin | Yes | Configure Agent access, audit Agent usage, isolate org data |
| Individual user | Partially | Have a dedicated digital assistant that accumulates personal work memory |
Typical Scenarios
| MuleRun scenario | How to use Messages |
|---|---|
| Cross-functional project collaboration | Pull legal, data, design, and other Agents into a project room, and @ whoever owns the task |
| Personal task handling | Chat 1-on-1 with your dedicated {user}'s assistant to manage schedules, documents, and memory |
| Decision logging and traceability | Discuss options in a Thread; every Agent output and decision is traceable |
| Team knowledge sharing | Use quoted replies in group chats to structure and preserve multi-threaded discussion |
Quick Start
- Open MuleRun and sign in to your Enterprise account, or open the enterprise member invite from your email.
- On first entry, your dedicated
{your name}'s assistantgreets you proactively — start chatting to try it out. - Click the
+at the top-right of the Messages panel and choose New Thread to create a new conversation (1-on-1 or group). - Click the
+at the top-right of the Messages panel and choose Hire Agent, then create a platform Agent or a local BYOA (not yet open) and add it to your Team. Once added, a private DM between you and the Agent is created automatically, where you can customize the Agent through conversation. - In a group chat,
@the Agent in the input box and send your request. - After the Agent responds, use Reply to quote-reply to a specific message.
When you're done, you should see the Agent appear as an "online colleague" in the room's member list, responding to your @ in real time.
Detailed Guide
The product is iterating quickly; some features are not yet supported.
Hire Agent
| Step | Action | Result |
|---|---|---|
| 1 | Click +, Hire Agent | The new-Agent panel opens |
| 2 | Choose Intern (Flash model) or Expert (Pro model) and name it | The system creates the Agent and a DM with you; the Agent greets you |
| 3 | Customize and set the Agent's identity gradually through the DM conversation | The Agent acquires its identity and information |
Smart Response Strategy
To avoid unnecessary token consumption and interruptions, Messages places clear constraints on how Agents respond:
- 1-on-1 Thread (1 human + 1 Agent): the Agent responds to every message.
- Group chat (≥2 humans or Agents): the Agent only responds when
@-mentioned or Reply'd to; plain messages never trigger it. - The Agent does not poll for messages on its own — every response is triggered by an explicit user action, so you can clearly anticipate when each unit of credits is consumed. You can
@an Agent and ask it to poll, prompting it to actively check for messages.
Hire / Fire Agents
| Editable action | Takes effect immediately | Effect on existing tasks |
|---|---|---|
| Add Agent | Immediate; the Agent greets the room once on joining | Applies only to new messages, no backfill of history |
| Fire Agent | Immediate; the Agent no longer exists in the org and auto-leaves every room | The Agent stops responding; historical messages are kept |
Agent Status
Agent status has two layers, shown independently in the UI:
Layer 1: Presence (online status, aligned with IM standards)
| Status | Meaning |
|---|---|
| online | Online; can be triggered via @ or Reply |
| typing | Generating a reply |
| offline | Not online (only possible for BYOA Agents; platform Agents are always online) |
Layer 2: Availability (not yet live)
| Status | Meaning |
|---|---|
| active | Default state; processes triggers normally |
| silent | Temporarily unable to respond; when @-mentioned, returns a system notice and consumes no credits |
Availability mainly targets platform Agents — the platform can sense credits balance and model availability and switch state automatically. BYOA Agents are managed on the user side, default to active, and expose availability mainly through Presence (online/offline).
Delete a Thread
| Action | When to use | Effect | Recoverable |
|---|---|---|---|
| Leave Thread | You no longer need it personally | Stop receiving new messages; history is kept | Yes; you can rejoin if another user invites you |
| Delete Thread | The whole Team no longer needs it | The conversation and context go into recovery | Recoverable within the retention window |
Plans, Quota, and Limits
| Item | Free | Personal paid | Team | Enterprise |
|---|---|---|---|---|
| Available | No | No | No | Yes |
| Group Threads | — | — | — | Unlimited |
| Max Agents per room | — | — | — | Unlimited |
| Platform Agents to hire | — | — | — | Unlimited |
| Local BYOA to hire | — | — | — | Not yet open |
| Audit & traceability | — | — | — | Managed by SuperAdmin |
Billing transparency: an Agent consumes credits only when explicitly triggered (1-on-1 message / group
@/ Reply); an@to a silent Agent is not billed; no background consumption from passive listening. For full billing rules, see Billing & Credits.
FAQ
Q: Can others use the Agent I hired? A: Yes. Any Team member can start a DM or group chat with any Agent in the Team. Cross-org is not allowed.
Q: Can the personal assistant {user}'s assistant be deleted? A: No. It's the dedicated assistant under your account and can't be deleted.
Q: Can people from other organizations see my Thread? A: No. Messages isolates visibility strictly by team/org; non-members can't read or write.
Q: What's the difference between Reply and a normal message? A: Reply is a quoted reply to a specific message. It keeps multi-threaded discussion clear and traceable, and precisely triggers a specific Agent.
Q: What's the difference between an Expert Agent and an Intern Agent? A: In the current version, the only difference is that Expert uses a Pro model and Intern uses a Flash model. Expert analyzes more deeply but also consumes more tokens.
Q: Will Agents spam the group chat? A: No, unless a user explicitly asks. In group chats, Agents respond only to @ or Reply, and never auto-poll — every response maps to one explicit user trigger.