Jarvis
A 100% private, offline AI voice assistant that runs locally on your computer with unlimited memory, MCP tool integration, and free dictation mode.
At a Glance
Free forever for personal use. Open-source, fully local, no subscriptions.
Engagement
Available On
Alternatives
Listed May 2026
About Jarvis
Jarvis is a fully local, privacy-first AI voice assistant that runs entirely on your computer without any cloud services or subscriptions. It understands natural conversation — say "Jarvis" anywhere in your sentence and it responds with context-aware, conversational replies. Jarvis maintains unlimited memory across sessions, supports thousands of external tools via MCP integration, and includes a free offline dictation mode as an alternative to cloud-based transcription services.
- 100% Local Processing — all AI inference runs on-device via Ollama; no data ever leaves your machine, and sensitive info like emails, tokens, and passwords is automatically redacted before being saved to disk.
- Conversational Awareness — maintains a rolling short-term context so Jarvis understands multi-person discussions and can answer follow-up questions like "Jarvis, what do you think?" without needing the full question repeated.
- Unlimited Memory — a self-organising knowledge graph learns from every conversation, auto-splits by topic, and surfaces relevant knowledge automatically; a Memory Viewer GUI lets you browse diary entries and the knowledge graph.
- Free Offline Dictation — hold a hotkey, speak, release to paste transcribed text into any app; supports filler-word removal, custom dictionaries, and a dictation history window, all without any cloud round-trip.
- MCP Integration — connect to thousands of external tools including Home Assistant, GitHub, Google Workspace, Slack, Discord, Notion, and databases; smart embedding-based tool selection prevents context rot as you add more tools.
- Built-in Tools — screenshot OCR, web search (DuckDuckGo → Brave → Wikipedia fallback chain), weather, file access, nutrition tracking, and location awareness are included out of the box.
- Natural Wake Word — say "Jarvis" anywhere in your sentence; interrupt with "stop"; follow up without repeating the wake word; echo detection prevents Jarvis from responding to its own speech.
- Adaptive Tone — automatically adjusts communication style to the topic: surgical for code, pragmatic for business, encouraging for wellbeing.
- Cross-Platform Desktop App — pre-built binaries for Windows, macOS, and Linux; a setup wizard guides model selection, Whisper configuration, and MCP server setup on first launch.
- Open Source — written in Python, source available on GitHub under a custom license (free for personal use; commercial use requires contact).
Community Discussions
Be the first to start a conversation about Jarvis
Share your experience with Jarvis, ask questions, or help others learn from your insights.
Pricing
Personal Use
Free forever for personal use. Open-source, fully local, no subscriptions.
- 100% local AI processing
- Unlimited memory
- MCP tool integration
- Free offline dictation
- Built-in web search, weather, OCR, nutrition tracking
Commercial Use
Commercial use requires contacting the developer for licensing.
- Commercial license for Jarvis
Capabilities
Key Features
- 100% local AI processing via Ollama
- Conversational context awareness across multi-person discussions
- Unlimited memory with knowledge graph and Memory Viewer GUI
- Free offline dictation mode (WisprFlow alternative)
- MCP integration for thousands of external tools
- Built-in web search with DuckDuckGo/Brave/Wikipedia fallback chain
- Screenshot OCR
- Weather and location awareness
- Nutrition tracking
- Natural wake word placement anywhere in sentence
- Echo detection
- Adaptive tone per topic
- Automatic redaction of sensitive info
- Filler word removal in dictation
- Custom dictation dictionary
- Task-list planner for multi-step queries
- Small-model digest passes to prevent context overflow
- Piper TTS (neural, local)
- Chatterbox TTS with voice cloning (source mode)
- CUDA acceleration for Windows speech recognition
- Setup wizard with graphical configuration
- Privacy hardening / fully offline mode
