AI Analysis
AI-powered insights from heatmaps, session data, and study results.
Pro AI Analysis features require a Pro license.
AI Providers
MouseViz supports 5 AI providers. Configure in Settings > AI:
| Provider | API Key | On-Device | Image Analysis |
|---|---|---|---|
| Apple Intelligence | Not needed | Yes | No |
| Anthropic (Claude) | Required | No | Yes |
| OpenAI (GPT) | Required | No | Yes |
| Ollama | Not needed | Yes (local) | Depends on model |
| OpenRouter | Required | No | Yes |
Apple Intelligence is free and requires no API key — it runs entirely on your Mac using the Foundation Models framework (Apple Silicon, macOS 26+). It works best for quick summaries and text-based analysis.
For image analysis (heatmaps) and complex tasks (study scripts, JSON generation), a cloud provider is required.
Auto-Insight
After every recording, MouseViz can automatically generate a 3-4 sentence summary highlighting:
- The most important finding
- Frustration patterns detected
- One actionable recommendation
Enable in Settings > AI > Auto-generate insight after recording.
When no API key is configured, Apple Intelligence provides the summary for free.
Heatmap Analysis
Send a heatmap image to AI for visual analysis. The AI identifies:
- Hot zones — high-activity areas and what they suggest
- Dead zones — areas with no interaction (discoverability issues)
- Frustration patterns — rage click clusters
- Movement inefficiencies — confusing navigation paths
- Actionable recommendations — specific UI improvements
This requires a cloud provider (Anthropic, OpenAI, or OpenRouter) since Apple Intelligence does not support image input.
Ask Your Data
Type natural language questions about your research data:
- "Which task had the highest failure rate?"
- "How do participants compare on the checkout task?"
- "What are the main frustration patterns?"
The AI answers using your actual session metrics, task results, and researcher notes.
Study Insights
After running a study with multiple participants, AI generates a comprehensive report:
- Executive summary — 2-3 sentences for stakeholders
- Key findings — with severity (critical/warning/info) and evidence
- Per-task recommendations — with priority (high/medium/low)
- Design changes — specific, actionable suggestions
- Next steps — what to test in the next iteration
Verify Connection
Click Verify in Settings > AI to test your provider connection before starting analysis.