Skip to Content
AgentsConfiguration Options

Configuration Options

Detailed guide to all agent configuration options.

Page Structure

The agent edit page has three tabs:

TabPurpose
ConfigurationCore settings (status, name, persona, LLM, messages)
Knowledge AccessControl which documents the agent can use
IntegrationsConnect to external channels (Telegram, Discord, etc.)

Configuration Tab

Status

Control your agent’s availability:

StatusPlaygroundIntegrationsNotes
Testing✅ Enabled❌ DisabledDefault for new agents. Test before going live.
Active✅ Enabled✅ EnabledReady for production use.
Inactive❌ Disabled❌ DisabledTemporarily disabled.
Archived❌ Disabled❌ DisabledHidden from main list.

If your agent has connected integrations and you change from Active to another status, you’ll see a confirmation dialog warning that the agent will stop responding on connected channels.

Basic Settings

  • Name — Identifies the agent in your dashboard
  • Description — Internal notes about the agent’s purpose

LLM Configuration

Control the AI model powering your agent.

Default Mode (Advanced toggle off):

  • Uses your organization’s tier defaults
  • Displays an info box explaining tier defaults are in use
  • Recommended for most use cases

Advanced Mode (Advanced toggle on):

SettingDescription
ModelSelect from available AI models based on your plan
TemperatureControls response creativity (0.0-2.0)

Temperature Guide

ValueBehavior
0.0 - 0.3Focused, deterministic responses
0.4 - 0.7Balanced creativity and consistency
0.8 - 2.0More creative, varied responses

Recommendation: Start with 0.3-0.5 for factual Q&A based on documents.

Show Document References

This toggle (always visible regardless of Advanced mode) controls whether the agent includes source citations in responses:

  • Off (default) — Responses don’t mention source documents
  • On — Responses include references to source documents

Enable document references for use cases requiring transparency, such as legal, compliance, or internal content where users need to verify sources.

Language Support

Your agent automatically detects the language of each user message and responds in kind — no manual configuration needed. 176 languages are supported out of the box, including Arabic, French, German, Spanish, Turkish, Chinese, Japanese, and more.

Agent Persona

The instructions that control agent behavior. See Agent Personas for best practices.

Actions available:

  • Generate — Opens the Agent Persona Builder
  • Edit — Opens a slide-out panel to write or modify the persona directly

Click on the persona text area to open the edit panel.

Platform Rules Box: Click to view the platform reliability rules that are automatically applied to all agents:

  • No fabrication — won’t invent facts
  • Acknowledges gaps — says “I don’t know” when appropriate
  • Resists speculation — won’t guess at missing information
  • No professional impersonation — won’t act as lawyers, doctors, etc.
  • Corrects false premises — points out incorrect assumptions
  • Blocks harmful content — refuses inappropriate requests

Platform rules ensure all agents behave reliably and safely. They cannot be disabled or modified.

Knowledge Access Tab

Control which documents your agent can access when responding to queries.

Access Modes

ModeBehaviorWhen to Use
Exclude ModeAgent accesses ALL documents EXCEPT those selectedHide sensitive documents while allowing access to everything else
Include ModeAgent accesses ONLY selected itemsRestrict agent to a specific set of documents

Selection Options

Fine-tune access using multiple selection types:

  • Folders — Select entire folders and all documents within them
  • Tags — Filter documents by their assigned tags
  • Individual Documents — Include or exclude specific documents within selected folders

Access Preview

The preview panel shows the effective document access based on your current selections. Use this to verify your configuration before saving.

Start with Exclude Mode if you want broad access with specific restrictions. Use Include Mode when you need tight control over exactly which documents the agent can reference.

Integrations Tab

Connect your agent to communication channels.

This tab shows available integration options for your agent. You can connect channels like:

  • Telegram
  • Discord
  • Slack
  • WhatsApp
  • Meta (Facebook Messenger)
  • And more (based on your plan)

See the Integrations documentation for detailed setup guides for each channel.

Human Handoff

Enable live chat handoff so your AI agent can transfer conversations to a real human when users request it.

How to enable:

  1. In the Configuration tab, scroll to the Human Handoff section
  2. Toggle Enable Human Handoff on
  3. Configure the messages:
    • Handoff Message — Shown to the user when the transfer starts
    • Accept Message — Auto-sent when a human agent accepts the conversation
    • Resolve Message — Shown when the human agent ends the conversation
    • Max Wait Time — Auto-expire if no agent accepts (5, 15, 30, 60, or 120 minutes; default: 30 minutes)
  4. Click Save Changes

When a user asks to speak with a human, the AI creates a handoff request visible on the Inbox page. A team member accepts, chats with the user on their channel, and resolves when done.

Human Handoff is available on all plans. Enable it per agent in the Configuration tab.

Testing Your Agent

Before saving changes, you can test your agent:

  1. Click the Test Agent button at the bottom of the form
  2. The Playground panel slides open
  3. Have a test conversation with your agent
  4. Close the Playground when done

The Test Agent button is disabled when the agent status is Archived or Inactive. Change the status to Active or Testing to enable testing.

Saving Changes

ButtonAction
Save ChangesSave all modifications
Test AgentOpen the Playground to test

To discard changes, navigate away using the back button or sidebar navigation.

What gets saved:

  • Only changed fields are updated
  • You’ll see “No changes to save” if nothing was modified
  • Success or error messages appear after save

Partial Save Handling: If some changes save but others fail, you’ll be notified which sections failed and the page will refresh to show the current state.

Changes to agent settings take effect immediately after saving. Test in the Playground after making changes.

Last updated on