Home › How-To › Starting a Conversation

Starting a Conversation

Create a bot, select an LLM model, and begin a multi-turn conversation in OpenRails

Overview

OpenRails Chat supports multiple LLM providers including local AI models, OpenAI GPT-4/5, Anthropic Claude, and Google Gemini. Every conversation belongs to a bot, which stores settings like model selection, system prompt, and linked data lakes for RAG context.

Chatbot Setup page

Create a Bot

Create New Chatbot wizard

Navigate to Bots

From the sidebar, click Bots to view the bot list for your current project.

Click "New Bot"

Click New Bot in the upper-right corner to open the bot creation form.

Configure Bot Settings

Fill in the required fields:

  • Bot Name — A descriptive name for this bot
  • LLM Model — Select from available models (GPT-4, Claude, Gemini, local models)
  • System Prompt — Instructions that define the bot's behavior and personality
  • Data Lakes — Optionally link data lakes for RAG context injection

Save the Bot

Click Save to create the bot. You will be taken to the conversation interface.

Start a Conversation

Open the Bot

Click on a bot from the bot list to open the conversation interface.

Type Your Message

Enter your message in the input field at the bottom of the chat panel. Press Enter to send or Shift+Enter for a new line.

View the Response

The LLM response streams in real time. If RAG is enabled, you will see citation indicators linking to the source documents used for context.

Continue the Conversation

OpenRails maintains full multi-turn conversation history. Each message is sent with the complete conversation context, allowing the LLM to reference earlier parts of the discussion.

Model Selection Reference

Selecting an AI model
Provider Models Best For
Local AI Models Llama 3, Mistral, Phi Privacy-sensitive tasks, offline usage, cost-free inference
OpenAI GPT-4, GPT-4o, GPT-5 General-purpose tasks, code generation, analysis
Anthropic Claude 3.5 Sonnet, Claude 4 Opus/Sonnet Long-form content, careful reasoning, document analysis
Google Gemini 1.5 Pro, Gemini 2.0/2.5 Multimodal tasks, large context windows
Tip: You can switch models mid-conversation by editing the bot settings. The conversation history is preserved, so the new model picks up where the previous one left off.
Important: Ensure the selected model's API key is configured before creating a bot. See Configuring LLM Keys for setup instructions.

Next Steps