RAG ChatsRAG Chats

Create Your First Bot

A bot is your AI assistant that answers questions using your documents. Each bot has its own knowledge base and configuration.

Creating a Bot

Navigate to the Dashboard

After logging in, you'll land on the dashboard. Click the + New Bot button in the top right corner.

Enter Bot Details

Fill in the basic information:

  • Name: A descriptive name like "Customer Support Bot"
  • Description: Optional description of what this bot does

Select AI Model

Choose the AI model that powers your bot. Each model has different capabilities and costs:

Configure Settings

Optionally adjust the advanced settings like temperature, max tokens, and system prompt. The defaults work well for most use cases.

Create Bot

Click Create Bot to finish. You'll be redirected to the bot dashboard where you can add data sources.

Available AI Models

Choose the model that best fits your needs. You can change this later in bot settings.

AI Models

GPT-4oOpenAIdefault: ✓

Best overall performance. Fast, accurate, and great at following instructions. Recommended for most use cases.

GPT-4 TurboOpenAI

Powerful model with 128K context window. Good for very long documents.

GPT-3.5 TurboOpenAI

Faster and cheaper than GPT-4. Good for simple Q&A where speed matters.

Claude 3 OpusAnthropic

Most capable Claude model. Excellent reasoning and nuanced responses.

Claude 3 SonnetAnthropic

Balanced performance and cost. Great for production workloads.

Claude 3 HaikuAnthropic

Fastest Claude model. Best for high-volume, simple queries.

Gemini ProGoogle

Google's multimodal model. Good general-purpose performance.

Bot Settings

These settings control how your bot generates responses. The defaults are optimized for most use cases.

LLM Settings

temperaturenumberdefault: 0.7

Controls randomness in responses. Lower values (0.1-0.3) for factual answers, higher (0.7-1.0) for creative responses.

max_tokensintegerdefault: 1000

Maximum length of generated response. Increase for longer answers, decrease for shorter.

RAG Settings

top_kintegerdefault: 5

Number of document chunks to retrieve. More chunks = more context but slower responses.

similarity_thresholdnumberdefault: 0.5

Minimum relevance score (0-1) for retrieved chunks. Higher = stricter matching.

Tip

Start with the default settings. You can always fine-tune them later based on the quality of responses you're getting.

System Prompt

The system prompt defines your bot's personality and behavior. A good system prompt includes:

  • Who the bot is (e.g., "You are a helpful customer support agent")
  • What topics it should focus on
  • How it should respond (tone, format)
  • What it should avoid doing

Example System Prompt

You are a helpful customer support assistant for ACME Corp. 

Your role is to:
- Answer questions about our products and services
- Help customers troubleshoot common issues
- Guide users through our documentation

Guidelines:
- Be friendly and professional
- If you're unsure, say so and offer to connect them with human support
- Always cite the specific documentation when possible
- Keep responses concise but complete

Next Steps

Now that you've created your bot, it's time to add some data for it to learn from.

Add Data Sources →