RAG ChatsRAG Chats

Bot Configuration

Fine-tune your bot's behavior with AI model selection, RAG settings, custom prompts, and response customization.

Configuration Categories

🤖 AI Models

Choose from GPT-4o, Claude 3, Gemini Pro, and configure model parameters

🔍 RAG Settings

Configure retrieval parameters, reranking, and search behavior

💬 System Prompts

Customize bot personality, tone, and response guidelines

📚 Source Citations

Configure how sources are displayed with responses

Quick Settings Overview

Model Settings

  • Model: The AI model powering your bot (GPT-4o, Claude 3, etc.)
  • Temperature: Controls response randomness (0 = deterministic, 2 = creative)
  • Max Tokens: Maximum response length

Retrieval Settings

  • Top K: Number of document chunks to retrieve
  • Similarity Threshold: Minimum relevance score for retrieved chunks
  • Reranking: Use Cohere to improve retrieval quality
  • Hybrid Search: Combine vector and keyword search

Response Behavior

  • Allow General Knowledge: Let bot answer outside its context
  • Show Sources: Display citations with responses
  • Source Display Mode: How sources appear (inline, footer, expandable)

Accessing Settings

  1. Go to your bot's dashboard
  2. Click Settings in the sidebar
  3. Navigate to the relevant section
  4. Make changes and click Save

API Configuration

All settings can be configured via the Bots API:

PATCH /api/agents/{agentId}/
{
  "model": "gpt-4o",
  "temperature": 0.7,
  "maxTokens": 1000,
  "topK": 5,
  "similarityThreshold": 0.5,
  "useReranking": true,
  "systemPrompt": "You are a helpful assistant...",
  "showSources": true
}

Best Practices

  • Start with defaults: The default settings work well for most use cases
  • Test incrementally: Change one setting at a time and test
  • Monitor performance: Use analytics to track how changes affect quality
  • Document changes: Keep notes on what configurations work best