🩺 Vitals
- 📦 Version: v0.56.0 (Released 2026-05-27)
- 🚀 Velocity: Active (Last commit 2026-05-29)
- 🌟 Community: 29.4k Stars · 2.8k Forks
- 🐞 Backlog: 117 Open Issues
🏗️ Profile
- Official: tinyhumans.ai/openhuman
- Source: github.com/tinyhumansai/openhuman
- License: GPL-3.0
- Deployment: Desktop | Local LLM (Ollama)
- Data Model: On-device SQLite + Obsidian-style Markdown vault
- Jurisdiction: UAE 🇦🇪 / Undisclosed (TinyHumans AI — founder-led)
- Compliance (SaaS): N/A (Undisclosed)
- Compliance (Self-Hosted): Self-Hosted (User Managed)
- Complexity: Low (1/5) - Native installers; runs as a standard desktop app
- Maintenance: Medium (3/5) - Fast-moving beta; single-maintainer continuity risk
- Enterprise Ready: Low (1/5) - Personal-use beta; no SSO, RBAC, audit, or certifications
1. The Executive Summary
What is it? OpenHuman is a local-first AI desktop assistant that builds a continuous, persistent memory of your work — a "second brain" — and stores it entirely on your own device in an Obsidian-compatible Markdown vault and a local SQLite database. It connects to your applications and can call tools, and it can run inference on-device via Ollama, so a working setup need never send your data to a vendor cloud.
The Strategic Verdict:
- 🔴 For Enterprise / Regulated Deployment: Hard Reject (today). This is a beta-stage, single-founder personal-productivity tool with no SSO, RBAC, audit trail, or certifications. Do not sanction it as a corporate system of record or feed regulated data into it.
- 🟢 For Privacy-Conscious Knowledge Workers: Strong Buy (Conditional). If you want a persistent AI assistant whose memory you fully own — portable plain-text Markdown, local SQLite, optional offline models — OpenHuman is a credible local-first alternative to cloud assistants that retain your history on their servers.
2. The "Hidden" Costs (TCO Analysis)
| Cost Component | ChatGPT (SaaS) | OpenHuman (Self-Hosted) |
|---|---|---|
| Pricing Model | Per-seat subscription (Plus / Team / Enterprise) | Free beta (BYO LLM keys or local models) |
| Memory & Data Location | Conversation history held on vendor servers | Local Markdown vault + SQLite on your device |
| Offline Capability | None — cloud-dependent | Optional, fully on-device via Ollama |
| Data Exit / Lock-in | Proprietary, export-limited memory | Plain-text Obsidian vault — no lock-in |
3. The "Day 2" Reality Check
🚀 Deployment & Operations
- Installation: Native installers for every desktop OS — Homebrew tap /
.dmg(macOS); signed apt repo,.deb,.AppImage, AUR (Linux);.msi(Windows). There is no server to provision or operate. - Scalability: Single-user desktop scope by design. There is no multi-user team server or central console — "scaling" means per-device installs, not centralized administration. This is a personal tool, not infrastructure.
🛡️ Security & Governance (Risk Assessment)
- Jurisdiction & Supply-Chain Trust: TinyHumans AI is a founder-led entity tied to a UAE (Dubai) jurisdiction with otherwise undisclosed corporate structure. Because OpenHuman executes locally and stores data on-device, jurisdictional data-residency exposure is largely moot — there is no vendor cloud holding your vault. The residual concern is not data location but supply-chain trust: you are accepting frequent updates from a single maintainer.
- The Compliance Shift (Endpoint Burden): As a local-first desktop app, OpenHuman holds no certifications and assumes no compliance liability. Your "second brain" is only as protected as the endpoint it lives on — full-disk encryption, device management, and backup are 100% your responsibility. There are no server-side controls because there is no server.
- License Risk (GPL-3.0 Copyleft): Released under GPL-3.0. For an individual running it as a standalone desktop assistant, the risk is minimal. For any team tempted to fork it or embed its components into proprietary internal software, the strong copyleft obligation applies — distribution would trigger source-disclosure requirements. Treat it as an application to use, not a codebase to quietly absorb.
4. Market Landscape
🏢 Proprietary Incumbents
- ChatGPT: The dominant cloud assistant, now with persistent memory, a desktop app, and app connectors — but your history and memory live on OpenAI's infrastructure, not yours.
- Rewind AI: A personal-memory assistant that records and recalls your digital activity, but proprietary and cloud-tied, with your timeline stored off-device.
🤝 Open Source Ecosystem
- Jan: A local-first desktop assistant focused on running LLMs entirely on your own hardware. It is the closest peer on privacy, but it is a model-running chat client — it does not build OpenHuman's continuous, Obsidian-style memory vault.
- AnythingLLM: A private, local desktop app for RAG and agents over your documents. It excels at chatting with a defined corpus, whereas OpenHuman's distinction is a continuously accreting personal memory rather than workspace-scoped document retrieval.