AI on your terms.
Meet Duncan — your private AI assistant. RAG-powered semantic search across all your data, intent classification that executes actions across plugins, chat-based interaction, multi-provider support (OpenAI, Groq, Mistral, local Llama), and speech-to-text. Everything runs on your device or through your own API keys. Every AI feature is opt-in.
AI that stays on your device
Duncan is PrivStack's AI assistant built for privacy. Semantic search finds answers across all your notes, tasks, journal entries, and more using RAG vector indexing — no cloud required. Tell Duncan what to do and it classifies your intent, then executes actions across plugins: create tasks, add events, draft notes. Choose from multiple AI providers or run models locally. Every feature is opt-in.
AI that phones home
AI features usually mean sending your data to cloud servers where it may be used to train models, analyzed by employees, or stored indefinitely. You're locked into one provider with no control.
Semantic Search & RAG
Ask Duncan a question and it searches across all your data using vector embeddings. Find related notes, tasks, calendar events, and journal entries by meaning, not just keywords.
- RAG vector indexing across all plugins
- Semantic search by meaning, not just keywords
- Cross-plugin search — notes, tasks, journal, calendar, contacts, and more
- Context-aware suggestions based on what you’re working on
Intent & Action Execution
Tell Duncan what to do in natural language. It classifies your intent and executes actions: create a task, add a calendar event, draft a note, log a habit.
- Natural language intent classification
- Action execution across plugins
- Create tasks, events, notes, habits, and more via chat
- Action result feedback — see what was created
Multiple AI Providers
Choose your AI provider: OpenAI, Groq, Mistral, or run models locally with Ollama (Llama, etc.). Your API keys stay on your device.
- OpenAI, Groq, and Mistral cloud providers
- Local models via Ollama (Llama and others)
- Speech-to-text via local Whisper model (Cmd+M)
- Bring your own API keys — we never see them
Ready to get started?
Try PrivStack free for 7 days. Your data stays on your devices.