explaingit

tinyhumansai/openhuman

★ 12,323RustAudience · generalComplexity · 3/5Setup · moderate

TLDR

A desktop AI personal assistant that connects to your email, calendar, and 100+ services, builds a private local knowledge base about you, and answers questions with your full context already loaded.

Mindmap

mindmap
  root((OpenHuman))
    What it does
      Personal knowledge base
      Account integration
      AI assistant
      Video call participation
    Tech Stack
      Rust
      Obsidian format
      AI model routing
    Use Cases
      Personal AI context
      Local data privacy
      Meeting assistance
    Audience
      Knowledge workers
      Privacy-conscious users
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Connect your Gmail, calendar, and documents so the AI can answer questions with your real personal context.

USE CASE 2

Build a private, local knowledge base from your accounts that you can also browse and edit in Obsidian.

USE CASE 3

Use the AI assistant in video calls as a participant that can look up your notes and context in real time.

USE CASE 4

Let the app automatically pick the right AI model for each task without manually choosing between providers.

Tech stack

Rust

Getting it running

Difficulty · moderate Time to first run · 30min

Early beta, requires downloading the desktop app and authorizing accounts via OAuth before the knowledge base can be built.

License not specified in the explanation.

In plain English

OpenHuman is an open-source personal AI assistant, a desktop application that connects to your existing tools and learns about you over time. Instead of starting from scratch every session, it continuously pulls information from your accounts (email, calendar, documents, project management tools, and more) and builds a local knowledge base on your computer. This means the AI already knows your context the next time you open it. The core idea is a "memory tree", a structured, locally stored summary of everything you connect. Your data is chunked into small pieces, scored for relevance, and saved as files on your machine in a format compatible with Obsidian (a popular note-taking app), so you can browse and edit it yourself. Nothing is sent to a remote server by default. The assistant has a desktop mascot, a visual face, that can speak, react, and even join video calls as a participant. It comes with built-in tools for web search, reading web pages, working with files and code, and voice input and output. A feature called TokenJuice compresses long content before sending it to an AI model, which the project claims reduces cost and response time significantly. It connects to over 100 services including Gmail, GitHub, Slack, and Google Drive using one-click authorization. Model routing automatically picks the right AI model for each task. The project is in early beta, built with Rust, and available for download on its website. The full README is longer than what was provided.

Copy-paste prompts

Prompt 1
I installed OpenHuman and connected my Gmail. How do I ask it to summarize emails from a specific sender from the past week?
Prompt 2
How does OpenHuman's memory tree work? Explain how it chunks and scores data from my accounts and saves it locally as files.
Prompt 3
I want OpenHuman to join a video call as a participant. How do I set up the desktop mascot and enable it for meetings?
Prompt 4
Show me how to connect GitHub and Slack to OpenHuman and what kinds of work questions I can then ask it.
Prompt 5
How does OpenHuman's TokenJuice feature compress long content before sending it to an AI model, and how do I enable or configure it?
Open on GitHub → Explain another repo

← tinyhumansai on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.