Agent Settings

Configure your AI agent's behavior, personality, and appearance

Agent Settings#

Customize how your AI agent behaves, responds, and appears to users.

Accessing Agent Settings#

1. Go to your agent dashboard
2. Click Settings in the sidebar
3. Navigate through the sections

Settings Overview#

General Settings#

Agent Name#

The name shown to users in chat. Keep it friendly and memorable.

Description#

Internal description for your team: what this agent is for, what channels it serves, any special notes. This is not shown to end users.

Avatar#

Upload a profile picture for your agent. Recommended: 256x256 pixels, PNG/JPG/WebP format, under 2MB.

AI Settings#

AI Model#

Choose which AI model powers your agent. Models are organized by credit cost:

• Budget models (1 credit) -- Fast responses, good for simple queries
• Mid-tier models (5 credits) -- Balanced speed and quality
• Premium models (10-15 credits) -- High quality for complex conversations
• Top-tier models (25 credits) -- Best quality, slower response times

Available models and their credit costs are shown in the model selector in your dashboard. Models and pricing are updated periodically as new options become available.

Market / Region#

Optimizes AI responses, intent detection, and templates for your target audience:

• Philippines -- Filipino/Taglish language support
• US -- English language
• Global -- Multilingual support

Timezone#

Used for scheduling, calendar bookings, and time displays. Defaults to Asia/Manila.

Currency#

Used for pricing displays, analytics, and CRM reporting. Select your business currency (PHP, USD, etc.) from the dropdown.

Temperature (Response Creativity)#

Controls the balance between creativity and consistency in responses:

• Automatic Optimization (default): Creativity is auto-adjusted based on conversation type. For example, factual questions get lower creativity while open-ended conversations get more.
• Custom: Manual slider from 0.0 to 0.7
  • 0.0-0.3: Consistent, factual, predictable
  • 0.4-0.6: Balanced
  • 0.7: Maximum allowed creativity

Lower values are better for customer support and factual queries. The maximum is capped at 0.7 to maintain response quality. Intent-based adjustments still apply even with custom settings.

Communication Style#

• Conversation Style: Tone, formality level, personalization
• Language Style: Preset patterns, adaptation level
• Both can be detected automatically from your training data

Customer Image Analysis#

Toggle this on to let the AI analyze images uploaded by customers during conversations. Each image analyzed costs 1 additional credit.

System Prompt#

The core instructions that shape your AI agent's personality, tone, knowledge boundaries, and edge case handling. Changes to the system prompt take effect after retraining.

Assistant Team#

Create specialized AI personas that handle different types of conversations. The system automatically routes messages to the right team member based on the conversation topic.

Creating a Team Member#

1. Click Add Team Member
2. Set a name and system prompt for the persona
3. Define routing rules (keywords or topics that trigger this persona)
4. Save

Example Team#

Each team member uses the same underlying AI model but with different system prompts, allowing specialized behavior without creating separate agents.

Availability Settings#

Business Hours#

Set when your AI should respond normally (e.g., Monday-Friday, 9:00 AM - 6:00 PM).

After-Hours Behavior#

Choose what happens outside business hours:

• Continue normally: AI responds as usual
• Modified response: Different greeting mentioning after-hours
• Collect info only: Take messages for follow-up during business hours

Holidays#

Set specific dates when special messaging applies, with custom holiday messages.

Security Settings#

Domain Whitelist#

Restrict the web chat widget to only load on your approved domains. Requests from other domains are blocked.

Rate Limiting#

Prevent abuse by setting per-user limits on messages per minute and messages per day.

CAPTCHA#

Enable reCAPTCHA v3 for anonymous widget users to block automated abuse.

Webhooks#

Configure real-time notifications for agent events:

1. Click Add Webhook
2. Enter your HTTPS endpoint URL
3. Select the events to subscribe to
4. Copy and securely store the signing secret
5. Save

Available Events#

• leads.submit -- Lead form completed
• message.sent -- AI sends a message
• message.received -- User sends a message
• conversation.started -- New conversation begins
• conversation.ended -- Conversation closes
• tool.executed -- AI action succeeds
• tool.failed -- AI action fails
• integration.connected -- Platform integration connected
• integration.disconnected -- Platform integration disconnected
• agent.enabled -- Agent activated
• agent.disabled -- Agent deactivated
• agent.trained -- Training completed
• agent.training_failed -- Training failed

See the Webhooks documentation for setup details and payload examples.

Business Profile and CRM#

CRM configuration is available in Settings > AI. See Contacts and CRM for details on:

• Business Profile: AI-detected business type with templates and characteristics
• Lead Scoring: Signals, grades, and scoring weights
• Follow-ups: Automatic follow-up configuration
• Contacts: Auto-classified contacts with sentiment and journey stages

When Settings Take Effect#

Always click Save Changes at the bottom of each settings page.

Best Practices#

1. Test after changes -- Use the Playground to verify your agent behaves as expected
2. Start conservative -- Begin with lower temperature and shorter responses, then adjust
3. Iterate gradually -- Make small changes and test before making more
4. Review regularly -- Re-evaluate settings monthly as your business needs evolve

Related Pages#

• Understanding Agents
• Training Your Agent
• Testing Your Agent
• Contacts and CRM