Enterprise Chat
Chat with your AI agents in real time — with @mentions for multi-agent conversations, inline actions, handoff timelines, and a feedback loop that continuously improves agent quality.
Starting Conversations
Click the New Chat button to start a conversation. You'll see an agent selection grid showing all 11 deployed agents, each with their role emoji:
| Agent | Icon | Agent | Icon |
|---|---|---|---|
| Sales | 💼 | HR | 👥 |
| Support | 🛠 | Legal | ⚖ |
| Marketing | 📣 | Product | 📋 |
| Finance | 💰 | Research | 🔍 |
| Operations | ⚙ | Strategy | 🎯 |
| CEO | 👑 |
Click an agent to start a one-on-one conversation, or select multiple agents for a group chat.
@Mentions
Type @ in the message composer to mention an agent. An autocomplete dropdown appears with available agents.
Keyboard Navigation
- Arrow Up / Down — Navigate through the agent list
- Enter or Tab — Select the highlighted agent
- Escape — Close the dropdown without selecting
Visual Treatment
Mentioned agents are highlighted in their role color within the message text. When a message contains an @mention, the mentioned agent is automatically brought into the conversation if not already present.
@Sales can you research Acme Corp? will route the message to the Sales agent, who will respond with prospect research based on your company's knowledge base.
Multi-Agent Conversations
Multiple agents can participate in a single conversation. The conversation header shows agent badges for each participant, along with the total message count.
How It Works
- Start a chat with one agent, then @mention others to bring them in
- Each agent's messages are color-coded by their role
- Agents can see the full conversation history and build on each other's responses
- The conversation header updates with badges for all active agents
Agent Handoff
When one agent determines that a request is better handled by another specialist, it initiates a handoff. The chat displays a visual handoff timeline showing the delegation chain.
Handoff Timeline
- Arrow indicators show the direction of delegation between agents
- Each handoff includes a reason (e.g., “Routing to Legal for contract review”)
- The receiving agent picks up with full context from the preceding conversation
- Multiple handoffs can chain together (e.g., Sales → Legal → Finance)
Inline Actions
Agent responses can include action buttons that let you execute operations directly from the chat.
Available Actions
| Action | What It Does |
|---|---|
| Approve | Approve a pending request (e.g., campaign draft, expense report) |
| Reject | Reject a pending request with optional reason |
| Create Mission | Create a new mission in the Command Center from the agent's recommendation |
| Create Workflow | Generate a workflow in the Workflow Center based on the conversation |
When an action is executed, the button shows a checkmark to confirm completion. The action result is logged in the conversation history and the audit trail.
Feedback System
Every agent response includes feedback controls that drive the continuous improvement loop.
Thumbs Up
A single click to indicate the response was helpful. No additional input required — the positive signal is recorded immediately.
Thumbs Down
Opens a detailed feedback form with two fields:
- Root cause — Select from categories: Incorrect information, Missing context, Wrong agent, Off-topic, Compliance issue
- Suggestion — Free-text field to describe what the correct response should have been
Search & History
All conversations are persisted and searchable.
Full-Text Search
- Search across all conversations by keyword or phrase
- Results show snippet previews with highlighted matches
- Filter by agent, date range, or conversation status
Conversation Management
- Pin — Pin important conversations to the top of the sidebar for quick access
- Archive — Move completed conversations to the archive to keep the sidebar clean
- Archived conversations remain searchable and can be restored at any time
Info Panel
Toggle the info panel on the right side of the chat to see conversation metadata at a glance.
Panel Contents
- Agents involved — List of all agents that participated, with their role badges
- Message count — Total messages in the conversation
- Feedback score — Aggregate thumbs-up vs. thumbs-down ratio
- Handoff timeline — Visual chain of agent handoffs with timestamps
Sandbox Testing
Test agents in a safe environment before deploying them to production.
How to Use
- Navigate to Agent Workforce and select an agent
- Open the Test Panel tab
- Conversations in the test panel run in sandbox mode — no production data is affected
- Test different prompts, knowledge configurations, and edge cases
- Review responses for accuracy and compliance before deployment