Your Private
Chief of Staff.

The old way: Pasting your sensitive company data into ChatGPT and hoping for the best.The MiniMe way: An open-weights local LLM that can chat with your entire digital footprint without a single byte leaving your machine.

Omniscient, yet completely air-gapped.

Ask natural language questions about your own life.

Flawless RAG Citations

Hallucinations are unacceptable for your personal data. Every UI answer cites its exact local sources—whether it was a Slack message you read yesterday or a PDF you scrolled through last week.

"You discussed the new auth flow with Mark [1] and drafted the spec in Figma [2]."
[1] Slack: #engineering (10:42 AM)
[2] Desktop: Auth_Flow.fig (1:15 PM)

Native Voice UI

Talk directly to MiniMe hands-free using Web Speech API dictation.

Proactive System Insights

You don't always have to ask. MiniMe observes your habits and sends unintrusive push notifications with high-value suggestions.

Flow State Warning

You've been deep in focus for 2.5 hours. Taking a break now will boost afternoon retention by 40%.

Meeting SummarizerActive Plugin
Email Drafter1-Click Install

Agentic Plugins

Your AI Copilot doesn't just answer questions; it takes action. Extend MiniMe's capabilities by installing local plugins to automate repetitive tasks based on your graph data.

BYO Intelligence.

Your private Chief of Staff is powered by state-of-the-art weights running directly on your hardware, connected to your semantic vector store.

Ollama Native

Seamlessly connect to Ollama to run Llama 3, Mistral, or WebLLM entirely locally ensuring absolute zero-trust execution for sensitive queries.

Qdrant Vector DB

Semantic search is powered by a local Qdrant instance, enabling lightning-fast Retrieval-Augmented Generation mapped directly back to Neo4j.

AES Encrypted API Keys

Need more reasoning power? Opt-in to use OpenAI or Anthropic directly. Your keys are stored locally and encrypted using AES-256-GCM.

Perfect for Information Hoarders.

Busy Executives"Summarize all my Slack messages and unread emails from the last 4 hours while I was in meetings."
Researchers"Which PDF contained the neural network diagram that I was reading yesterday morning?"
Engineers"What file was I editing when I encountered that WebGL rendering error right before lunch?"
Writers"Pull up the character notes doc I briefly opened last week about the main antagonist."

Ready to run it locally?

Discover the hyper-optimized Rust architecture that powers the Desktop Guardian tracking agent.

Explore Desktop Architecture