Setting Up AI Providers

InsightLens supports multiple AI providers to give you flexible options for natural language analysis. This guide walks you through setting up and configuring different AI services.

Supported AI Providers

OpenAI (ChatGPT)

Models: GPT-4o, GPT-4o Mini, GPT-4 Turbo, GPT-3.5 Turbo
Best For: General analysis, reliable responses, wide language support
Cost: Pay-per-use via OpenAI API
Setup Difficulty: Easy

Anthropic (Claude)

Models: Claude 3.5 Sonnet, Claude 3.5 Haiku, Claude 3 Opus
Best For: Detailed analysis, safety-focused responses, complex reasoning
Cost: Pay-per-use via Anthropic API
Setup Difficulty: Easy

Local Models (Ollama)

Models: Llama 3.2, Llama 3.1, Mistral, Qwen 2.5 Coder
Best For: Privacy, offline use, no usage costs after setup
Cost: Free (hardware requirements apply)
Setup Difficulty: Moderate

Custom OpenAI-Compatible APIs

Providers: LM Studio, LocalAI, Text Generation Web UI
Best For: Self-hosted solutions, custom models
Cost: Varies by provider
Setup Difficulty: Advanced

Getting API Keys

OpenAI API Key

Visit OpenAI’s website
Create account or sign in to existing account
Navigate to API Keys section
Create new secret key
Copy and save the key securely (starts with sk-)

Important Notes:

Keys are only shown once - save immediately
Keep keys secure and never share publicly
Monitor usage to control costs
Set usage limits to prevent unexpected charges

Anthropic API Key

Visit Anthropic’s Console
Create account or sign in
Go to API Keys section
Generate new key
Copy and save the key securely (starts with sk-ant-)

Billing Setup:

Credit card required for API access
Pay-as-you-go pricing model
Usage monitoring tools available
Set spending limits for cost control

Configuration in InsightLens

Basic Setup Steps

Open InsightLens and navigate to Settings (⚙️ in sidebar)
Locate the “AI Assistant” section
Select your preferred provider from dropdown
Enter your API key in the secure field
Choose your preferred model
Test connection to verify setup

Provider-Specific Configuration

OpenAI Setup

API Provider: OpenAI
API URL: https://api.openai.com (auto-filled)
API Key: sk-your-openai-key-here
Model: gpt-4o-mini (recommended for most users)

Model Selection Guide:

GPT-4o Mini: Best balance of cost and performance
GPT-4o: Higher quality, more expensive
GPT-4 Turbo: Good for complex analysis
GPT-3.5 Turbo: Budget option, lower quality

Anthropic (Claude) Setup

API Provider: Claude (Anthropic)
API URL: https://api.anthropic.com (auto-filled)
API Key: sk-ant-your-anthropic-key-here
Model: claude-3-5-sonnet-20241022 (recommended)

Model Comparison:

Claude 3.5 Sonnet: Best overall performance
Claude 3.5 Haiku: Faster, less expensive
Claude 3 Opus: Most capable, highest cost

Ollama (Local) Setup

API Provider: Ollama (Local)
API URL: http://localhost:11434 (default)
Model: llama3.2 (or your installed model)

Prerequisites:

Ollama installed and running on your system
At least one model downloaded
Sufficient RAM (8GB+ recommended)

Local AI Setup (Ollama)

Installing Ollama

Windows

Download Ollama installer from ollama.ai
Run installer and follow setup wizard
Open Command Prompt or PowerShell
Verify installation: ollama --version

macOS

Download Ollama for macOS from ollama.ai
Install the application
Open Terminal
Verify installation: ollama --version

Linux

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Verify installation
ollama --version

Installing Models

Recommended Models for Survey Analysis

Llama 3.2 (3B) - Fast, good for basic analysis

ollama pull llama3.2

Llama 3.1 (8B) - Better quality, moderate resource usage

ollama pull llama3.1

Mistral (7B) - Good alternative with different strengths

ollama pull mistral

Qwen 2.5 Coder (7B) - Excellent for structured data analysis

ollama pull qwen2.5-coder

Running Ollama

Starting the Service

# Start Ollama (usually starts automatically)
ollama serve

# Test with a simple query
ollama run llama3.2 "Hello, how are you?"

Checking Available Models

# List installed models
ollama list

# Check running models
ollama ps

System Requirements for Local Models

Minimum Requirements

RAM: 8GB (for 3B models)
Storage: 4GB per model
CPU: Modern multi-core processor
OS: Windows 10+, macOS 10.15+, Linux

Recommended Specifications

RAM: 16GB+ (for 7B+ models)
GPU: Optional but improves performance
Storage: SSD for better model loading
CPU: Recent Intel/AMD with good single-thread performance

Environment Variables (Optional)

For enhanced security and convenience, you can set API keys as environment variables:

Windows

# Command Prompt
set OPENAI_API_KEY=sk-your-key-here
set ANTHROPIC_API_KEY=sk-ant-your-key-here

# PowerShell
$env:OPENAI_API_KEY = "sk-your-key-here"
$env:ANTHROPIC_API_KEY = "sk-ant-your-key-here"

macOS/Linux

# Add to ~/.bashrc or ~/.zshrc
export OPENAI_API_KEY="sk-your-key-here"
export ANTHROPIC_API_KEY="sk-ant-your-key-here"

# Apply changes
source ~/.bashrc

Benefits of Environment Variables

Enhanced Security: Keys not stored in application settings
Easy Management: Change keys without opening InsightLens
Team Sharing: Share configuration without exposing keys
Backup Safety: Keys not included in setting backups

Testing Your Setup

Connection Test

In Settings, click “Test Connection” button
Wait for response - should show green checkmark
Check error messages if test fails
Verify model availability in dropdown

First AI Query

Navigate to “Ask InsightLens” page
Enter simple question: “How many units are in my database?”
Wait for response and generated chart
Verify results match your expectations

Troubleshooting Connection Issues

OpenAI Connection Problems

Invalid API Key: Check key format (starts with sk-)
Quota Exceeded: Check billing and usage limits
Rate Limiting: Wait a few minutes and retry
Model Availability: Ensure selected model is accessible

Anthropic Connection Problems

Authentication Failed: Verify API key accuracy
Insufficient Credits: Check account balance
Model Access: Ensure you have access to selected model
Regional Restrictions: Check if service is available in your region

Ollama Connection Problems

Service Not Running: Start Ollama service
Model Not Found: Download required model
Port Conflicts: Check if port 11434 is available
Memory Issues: Ensure sufficient RAM for model

Cost Management

Understanding Pricing

OpenAI Pricing (approximate)

GPT-4o Mini: $0.15/1M input tokens, $0.60/1M output tokens
GPT-4o: $5.00/1M input tokens, $15.00/1M output tokens
GPT-4 Turbo: $10.00/1M input tokens, $30.00/1M output tokens

Anthropic Pricing (approximate)

Claude 3.5 Sonnet: $3.00/1M input tokens, $15.00/1M output tokens
Claude 3.5 Haiku: $0.25/1M input tokens, $1.25/1M output tokens
Claude 3 Opus: $15.00/1M input tokens, $75.00/1M output tokens

Cost Optimization Tips

Choose Appropriate Models

Use cheaper models for simple queries
Reserve premium models for complex analysis
Test with small datasets first

Efficient Query Design

Be specific and concise in questions
Avoid unnecessary follow-up queries
Use filters to reduce data scope

Monitor Usage

Check API provider dashboards regularly
Set up billing alerts
Track usage patterns in InsightLens

Security Best Practices

API Key Security

Never share API keys publicly
Use environment variables when possible
Rotate keys periodically
Revoke unused or compromised keys

Data Privacy

Understand what data is sent to AI providers
Review privacy policies of your chosen provider
Consider local models for sensitive data
Implement data minimization strategies

Access Control

Limit who can configure AI settings
Use separate keys for different team members
Monitor usage for unusual patterns
Implement approval processes for new providers

Next: Asking Questions