LuceData

Add a New AI Engine

Add a New AI Engine

LuceData's AI Assistant requires you to Bring Your Own Model (BYOM) during the beta period. This guide will help you configure AI engines to power natural language to SQL generation.

Supported AI Providers

LuceData supports multiple AI providers:

ProviderTypeCostLocal/Cloud
OpenAIGPT-5, GPT-4+Pay-per-useCloud
Azure OpenAIGPT-5, GPT-4+Pay-per-useCloud
Anthropic ClaudeClaude 3 (Opus, Sonnet, Haiku)Pay-per-useCloud
Google GeminiGemini Pro, UltraPay-per-useCloud
OllamaLlama 3, Mistral, CodeLlama, etc.FreeLocal

💡 Tip: For the best SQL generation quality, we recommend GPT-5 or Claude 4.5 Sonnet. For a free option, try Ollama with Llama 3 or CodeLlama.

Before You Begin

Obtain an API Key

Choose your provider and get an API key:

OpenAI

  1. Visit platform.openai.com
  2. Sign up or log in
  3. Go to API Keys in your account settings
  4. Click "Create new secret key"
  5. Copy the key (starts with sk-...)
  6. Add credits to your account if needed

Azure OpenAI

  1. Log in to Azure Portal
  2. Create an Azure OpenAI resource (if you don't have one)
  3. Go to Keys and Endpoint
  4. Copy the Key and Endpoint URL
  5. Note your Deployment Name

Anthropic Claude

  1. Visit console.anthropic.com
  2. Sign up or log in
  3. Go to API Keys
  4. Click "Create Key"
  5. Copy the key (starts with sk-ant-...)

Google Gemini

  1. Visit ai.google.dev
  2. Sign up or log in
  3. Click "Get API Key"
  4. Create a new project or select existing
  5. Copy the API key

Ollama (Local, Free)

  1. Visit ollama.ai
  2. Download and install Ollama for your OS
  3. Open Terminal/Command Prompt
  4. Pull a model: 
    ollama pull llama3
# or
ollama pull codellama
# or
ollama pull mistral
  5. Start Ollama (it runs locally on http://localhost:11434)

Adding an AI Engine

Method 1: During First Launch

When you first open LuceData, you'll be prompted to configure an AI engine:

  1. Select Provider: Choose from the dropdown (OpenAI, Azure OpenAI, Claude, Gemini, Ollama)
  2. Enter Details: Fill in the required fields (see provider-specific instructions below)
  3. Test Connection: Click "Test" to verify the configuration
  4. Save: Click "Save" to complete setup

Method 2: From Settings

To add additional AI engines or modify existing ones:

  1. Click the Settings icon (⚙️) in the bottom-left corner
  2. Select "AI Engines" from the left menu
  3. Click "+ Add New Engine"
  4. Follow the configuration steps below

Configuration by Provider

OpenAI Configuration

Fields:

  • Engine Name: Give it a descriptive name (e.g., "OpenAI GPT-4")
  • API Key: Your OpenAI API key (starts with sk-...)
  • Model: Select from dropdown:
    • gpt-4 (recommended for quality)
    • gpt-4-turbo
    • gpt-3.5-turbo (faster, cheaper)
  • Organization ID (optional): If you belong to multiple organizations
  • Temperature: 0.0 - 1.0 (default: 0.7)
    • Lower = more deterministic
    • Higher = more creative

Example:

Name: OpenAI GPT-4
API Key: sk-proj-xxxxxxxxxxxxxxxxxxxxx
Model: gpt-4-turbo
Temperature: 0.7

Azure OpenAI Configuration

Fields:

  • Engine Name: Descriptive name (e.g., "Azure GPT-4")
  • Endpoint: Your Azure OpenAI endpoint URL
    • Example: https://your-resource.openai.azure.com/
  • API Key: Your Azure OpenAI key
  • Deployment Name: The name of your deployed model
  • API Version: Usually 2023-05-15 or 2024-02-15-preview
  • Temperature: 0.0 - 1.0 (default: 0.7)

Example:

Name: Azure GPT-4
Endpoint: https://my-openai.openai.azure.com/
API Key: abc123def456...
Deployment Name: gpt-4-deployment
API Version: 2024-02-15-preview
Temperature: 0.7

Anthropic Claude Configuration

Fields:

  • Engine Name: Descriptive name (e.g., "Claude 3 Opus")
  • API Key: Your Anthropic API key (starts with sk-ant-...)
  • Model: Select from dropdown:
    • claude-3-opus-20240229 (most capable)
    • claude-3-sonnet-20240229 (balanced)
    • claude-3-haiku-20240307 (fastest, cheapest)
  • Max Tokens: Maximum response length (default: 4096)
  • Temperature: 0.0 - 1.0 (default: 0.7)

Example:

Name: Claude 3 Opus
API Key: sk-ant-xxxxxxxxxxxxxxxxxxxxx
Model: claude-3-opus-20240229
Max Tokens: 4096
Temperature: 0.7

Google Gemini Configuration

Fields:

  • Engine Name: Descriptive name (e.g., "Gemini Pro")
  • API Key: Your Google AI API key
  • Model: Select from dropdown:
    • gemini-pro (text and reasoning)
    • gemini-ultra (most capable, limited access)
  • Temperature: 0.0 - 1.0 (default: 0.7)

Example:

Name: Gemini Pro
API Key: AIzaSyxxxxxxxxxxxxxxxxxxxxx
Model: gemini-pro
Temperature: 0.7

Ollama Configuration (Local)

Fields:

  • Engine Name: Descriptive name (e.g., "Ollama Llama 3")
  • Base URL: Usually http://localhost:11434
  • Model: The model you've pulled:
    • llama3 (recommended for SQL)
    • codellama (code-focused)
    • mistral
    • phi3
  • Temperature: 0.0 - 1.0 (default: 0.7)

Example:

Name: Ollama Llama 3
Base URL: http://localhost:11434
Model: llama3
Temperature: 0.7

💡 Tip: Make sure Ollama is running before testing the connection.

Testing Your Configuration

After entering your details:

  1. Click the "Test Connection" button
  2. LuceData will send a test request to the AI provider
  3. You'll see one of these results:
    • ✅ Success: Configuration is correct
    • ❌ Error: Review the error message and adjust your settings

Common errors:

  • Invalid API Key: Double-check your key is copied correctly
  • Network Error: Check your internet connection (for cloud providers)
  • Model Not Found: Verify the model name is correct
  • Rate Limit: You may have exceeded your provider's rate limits
  • Insufficient Credits: Add credits to your provider account

Managing Multiple AI Engines

You can configure multiple AI engines and switch between them:

Adding Multiple Engines

  1. Go to Settings → AI Engines
  2. Click "+ Add New Engine" for each provider you want to use
  3. Configure each engine with its own name and credentials

Switching Between Engines

  1. In the AI Assistant panel, look for the engine selector dropdown (top-right)
  2. Click the dropdown to see all configured engines
  3. Select the engine you want to use
  4. The selected engine will be used for all subsequent AI queries

Example Use Cases:

  • Use GPT-4 for complex queries requiring high accuracy
  • Use GPT-3.5 for quick, simple queries to save costs
  • Use Ollama for offline work or privacy-sensitive queries
  • Use Claude for queries requiring large context windows

Setting a Default Engine

  1. Go to Settings → AI Engines
  2. Find your preferred engine in the list
  3. Click the "Set as Default" button
  4. This engine will be selected automatically when you start LuceData

Editing an Existing Engine

  1. Go to Settings → AI Engines
  2. Find the engine you want to edit
  3. Click the Edit icon (pencil)
  4. Modify the settings
  5. Click "Test Connection" to verify
  6. Click "Save"

Deleting an Engine

  1. Go to Settings → AI Engines
  2. Find the engine you want to remove
  3. Click the Delete icon (trash can)
  4. Confirm the deletion

⚠️ Warning: You cannot delete the last remaining AI engine. At least one engine must be configured.

Best Practices

Security

  • ✅ Never share your API keys with others
  • ✅ Keep your API keys secure and rotate them periodically
  • ✅ Use environment-specific keys (dev vs. production)
  • ✅ Monitor your API usage and costs regularly

Performance

  • 💡 Use GPT-4 or Claude Opus for complex database schemas
  • 💡 Use GPT-3.5 or Claude Haiku for simple queries to reduce latency and cost
  • 💡 Use Ollama for local, offline work

Cost Management

  • 💰 Start with less expensive models (GPT-3.5, Claude Haiku)
  • 💰 Set up billing alerts with your AI provider
  • 💰 Use Ollama (free) for development and testing
  • 💰 Monitor your token usage in your provider's dashboard

Troubleshooting

"Invalid API Key" Error

  • Verify you copied the entire key without extra spaces
  • Check that the key hasn't been revoked or expired
  • Ensure you're using the correct key format for the provider

"Model Not Available" Error

  • Verify the model name is spelled correctly
  • Check that you have access to the model (some require waitlist approval)
  • For Azure, ensure your deployment name matches exactly

Connection Timeout

  • Check your internet connection
  • For Ollama, ensure the Ollama service is running
  • Try increasing the timeout in Settings → Advanced

High Costs

  • Switch to a cheaper model (GPT-3.5, Claude Haiku)
  • Use Ollama for routine queries
  • Review your usage patterns in the provider's dashboard

Next Steps

Now that you've configured your AI engine:

  1. ✅ Add a New Connection - Connect to a database
  2. ✅ Using the AI Assistant - Learn how to use natural language queries
  3. ✅ Using the Work Area - Write and execute SQL queries

Future: Commercial AI Model

In the commercial version of LuceData, we'll provide:

  • 🎯 Custom database-specialized AI model
  • 🔄 Updated weekly with the latest SQL best practices
  • 🚀 Optimized for database operations (DDL, DML, complex queries)
  • 💡 Pre-trained on common database patterns

BYOM will remain available as an option for users who prefer their own models.

Need help? Contact us at support@lucedata.com