Messages

Messages is the AI-native team chat built into MuleRun Enterprise. Unlike a normal group chat, Agents are first-class members here — you can @ them, Reply to them, and hire them into rooms. Long-running context lives in the Thread instead of starting from scratch in a new task tool every time.

Beta feature (Enterprise only)

Messages is currently in Beta and is available only on the Enterprise edition. Personal, Team, and Free plans are not eligible.

What It's For

• Keep humans and Agents in the same room with shared context, so the Agent doesn't need re-briefing on every task.
• Give an Agent a persistent identity for long-term work — research, writing, ops, support — instead of one-off task tools.
• Run multiple Agents side by side in one room with clear roles (e.g. "Researcher + Editor", "Support + Analyst").
• Use a private 1-on-1 with your personal assistant for everyday tasks and drafts.

Quick Start

1. Open the Messages panel in your Enterprise workspace. You'll see {Your Name}'s assistant on the left — a built-in 1-on-1 Thread that can't be deleted.
2. Click the + at the top-right and choose New Thread to start a 1-on-1 or group Thread. Pick the members (humans or Agents) and you're in.
3. Click the + at the top-right and choose Hire Agent to bring in a new Agent. Pick Intern (Flash model — lighter, cheaper) or Expert (Pro model — deeper analysis, higher cost), then give it a name.
4. After hiring, the system opens a private DM between you and the new Agent. Tell it who it is, what it should do, and the style you want — normal conversation is enough to customize it.
5. Bring an Agent into any group Thread by @-mentioning it, or Reply to one of its messages to keep it focused on that thread of discussion.

Threads, Agents, and How They Reply

Agents follow a strict set of response rules in Messages. Reading this section is the key to understanding the billing too:

• 1-on-1 Threads (1 human + 1 Agent): the Agent replies to every message you send.
• Group Threads (2 or more members, human or Agent): an Agent only replies when it's explicitly @-mentioned or Reply'd to. Plain group chatter never triggers it, so the room stays clean.
• Agents never auto-poll. Every Agent reply maps to one explicit user trigger. If you want an Agent to follow up on a schedule, @ it and tell it to — for example, "check this every 10 minutes and ping me on changes".

Reply (quoted reply) matters in groups: it makes parallel discussions traceable, and it lets you target one specific Agent without pulling in everyone else.

Hire and Fire Agents

Hire from Messages panel → top-right + → Hire Agent. Two tiers:

• Intern: Flash model. Faster responses, lower per-message cost. Good for everyday Q&A, summarizing, and light analysis.
• Expert: Pro model. Deeper reasoning and more complete answers, but higher per-message cost. Good for in-depth research, long-form writing, and complex analysis.

After hiring:

• Effect is immediate. When you add an Agent to a group it greets the room once — it does not backfill historical messages. It only sees the room from the moment it joined.
• Firing is also immediate. A fired Agent leaves every group it was in across the org. History is kept; the Agent simply stops responding.

BYOA (bring-your-own-agent) is not yet open — only the platform-provided Intern and Expert are available today.

Agent Status

Every Agent has a Presence indicator:

• online: ready, can be triggered via @ or Reply.
• typing: actively generating a reply, text will appear shortly.
• offline: only possible for future bring-your-own-agent setups — platform Agents are always online.

A future silent state will let an @ to that Agent receive a system message at zero cost, instead of a billed reply.

Threads — Leave vs Delete

• Leave: you step out of the Thread and stop receiving new messages. History is kept, and you can be re-invited.
• Delete: the Thread is removed for the whole team and moved into recovery. It can be restored within the retention window.

Delete is a team-wide action — confirm before removing.

Billing and Credits

Messages bills strictly on explicit triggers — there's no background polling cost:

• In a 1-on-1 Thread, every message you send produces one Agent reply and consumes Credits.
• In a group Thread, only an @ or Reply triggers the Agent. Plain group messages cost nothing.
• Intern (Flash) and Expert (Pro) consume Credits at different rates based on the tier you chose.
• The upcoming silent state will let an @ to that Agent return a zero-cost system message.

See Billing & Credits for the full rules.

FAQ

Can my teammates use the Agent I hired? Yes. Anyone in the same Team can DM or group-chat any Team Agent. Cross-org is blocked.

Can I delete my personal assistant? No. {Your Name}'s assistant is the default 1-on-1 Thread and is not deletable.

Can other organizations see my Thread? No. Threads are strictly isolated within your team and org.

What's the difference between Reply and a normal message? Reply is a quoted reply. It makes multi-thread discussion traceable and lets you point at one specific Agent without pulling in unrelated members.

Expert vs Intern — how do I choose? Expert uses a Pro model: deeper, longer answers, higher cost. Intern uses a Flash model: lighter, cheaper, good for everyday work. Start with Intern to validate the use case and upgrade only when you need more depth.

Will Agents spam the group? No. In groups, Agents only respond to @ or Reply, never auto-poll, and never react to unrelated chatter.