# Intelligence Settings

The Intelligence settings pane is where you configure your AI providers. Open it from Settings (Cmd+Comma) and select the Intelligence tab.

# Overview

The main view shows your available AI providers with their current status:

  • Claude in Weavestream — Cloud-based AI via Anthropic
  • Ollama in Weavestream — Local AI running on your Mac

Each provider shows:

  • An Active badge (green) if it's your current provider
  • A Set as Active button if it's configured but not selected
  • Turn On if it needs to be configured

A privacy link at the top provides more information about how data is handled.

# Configuring Claude

Click Claude in Weavestream to open its configuration:

# Status

Shows whether Claude is Configured (green checkmark) or Not Configured (red X).

# API Key

Enter your Anthropic API key. Keys start with sk-ant-. Click Save API Key to store it securely in your Mac's Keychain.

If you need an API key, click the link to visit Anthropic's website and create one.

# Removing the Key

If you want to disconnect Claude, click Remove Key. This deletes the API key from your Keychain. Your conversations are preserved — you just won't be able to send new messages until you add a key again.

# Configuring Ollama

Click Ollama in Weavestream to open its configuration:

# Status

Shows:

  • Configured / Not Configured — Whether settings have been saved
  • Connected / Cannot connect — Whether Weavestream can reach the Ollama server

# Server URL

The URL where Ollama is running. The default is http://localhost:11434, which is correct if Ollama is installed on the same Mac.

Click Test to verify the connection.

# Model

Select which Ollama model to use. If Weavestream detects running models, they appear in a dropdown. Click Refresh to re-discover available models.

If no models appear, make sure you've downloaded at least one model using Ollama (e.g., ollama pull llama3 in Terminal).

# Max Context Tokens

Controls the maximum context window size for the local model. Adjustable from 4,096 to 32,768 tokens in steps of 1,024.

  • Lower values — Faster processing, less data sent per request
  • Higher values — More data context, but slower and requires more memory

The default is suitable for most models. Increase it if you have a powerful Mac and want the AI to see more data at once.

# Experimental Notice

Ollama support comes with a note that local models may return less consistent results compared to Claude. Quality depends heavily on which model you choose and its size.

# Setup Instructions

If you haven't installed Ollama yet, the settings page includes a link to ollama.ai with setup instructions.

Click Save Configuration when you're done.

# Switching Providers

To switch your active AI provider:

  1. Make sure the provider you want to use is configured
  2. Click Set as Active next to that provider
  3. The green "Active" badge moves to your selection

You can switch at any time. Your conversations are preserved regardless of which provider is active.