AI Providers

Configure Anthropic Claude, OpenAI, or Local AI for Jack's responses.

Jack supports multiple AI providers. Choose based on your needs for quality, privacy, and cost.

Anthropic Claude

Claude excels at nuanced conversations and understanding context. Recommended for most deployments.

Setup

  1. Create an account at console.anthropic.com
  2. Go to API Keys and create a new key
  3. Copy the key (you won't see it again)
  4. In Jack's dashboard, go to Settings → Apps
  5. Find Anthropic and click Configure
  6. Paste your API key and save

Available Models

Model Best For Cost
Claude 3.5 Sonnet Complex reasoning, nuanced responses $3/$15 per 1M tokens
Claude 3 Haiku Fast responses, high volume $0.25/$1.25 per 1M tokens

OpenAI GPT

Fast and reliable with extensive language support.

Setup

  1. Create an account at platform.openai.com
  2. Go to API Keys and create a new secret key
  3. Copy the key
  4. In Jack's dashboard, go to Settings → Apps
  5. Find OpenAI and click Configure
  6. Paste your API key and save

Available Models

Model Best For Cost
GPT-4o Best quality, multimodal $5/$15 per 1M tokens
GPT-4o-mini Good balance, lower cost $0.15/$0.60 per 1M tokens

Local AI

Run AI entirely on your server. No API costs, complete privacy, works offline.

Setup

Local AI is enabled by default. No configuration needed.

Jack uses Transformers.js with the Xenova/all-MiniLM-L6-v2 model for embeddings. For text generation, it uses a lightweight model that runs in Node.js.

Considerations

  • Quality — Responses are simpler than cloud AI
  • Speed — First response may be slow as models load
  • Resources — Requires more RAM (~2GB for models)
  • Privacy — All processing stays on your server

Tip: Local AI works best for simple questions with clear answers in your knowledge base. For complex conversations, consider a cloud provider.

Ollama (Self-Hosted)

Run powerful open-source models locally with better quality than Transformers.js.

Setup

  1. Install Ollama from ollama.ai
  2. Pull a model: ollama pull llama3
  3. Start Ollama: ollama serve
  4. In Jack's dashboard, go to Settings → Apps
  5. Find Ollama and click Configure
  6. Enter the Ollama server URL (default: http://localhost:11434)
  7. Select your model and save

Recommended Models

Model Size RAM Needed
llama3:8b 4.7 GB 8 GB
mistral:7b 4.1 GB 8 GB
phi3:mini 2.3 GB 4 GB

Switching Providers

You can switch AI providers anytime:

  1. Go to Settings → Apps
  2. Configure the new provider with credentials
  3. Go to Settings → AI
  4. Select the new provider as active

Existing conversations continue seamlessly with the new provider.

Cost Optimization

  • Use smaller models — GPT-4o-mini or Claude Haiku for routine questions
  • Optimize knowledge base — Better retrieval means shorter prompts
  • Set autonomy to "Suggest" — Review before sending reduces unnecessary API calls
  • Use Local AI for simple queries — Route only complex questions to cloud AI