OpenHuman

• Asking for a morning briefing of what changed across Gmail, Slack, and Calendar in the last 12 hours
• Querying everything a colleague said about a project across email and chat
• Having the agent draft a workflow that summarizes new customer emails and posts them to Slack, then reviewing and saving it
• Sending an agent into a Google Meet call to transcribe it into memory and answer questions afterward
• Running a private assistant on a local Ollama model with Privacy Mode set to local-only
• Letting the agent tidy and rename a project folder inside a boundary you set, with each change approved

OpenHuman's pitch is that a personal agent is only as useful as what it already knows, so it does the ingestion for you: connect Gmail and Slack once and a local, readable memory fills itself every 20 minutes. That, plus a real desktop UI, agent-drafted workflows you approve on a canvas, and a default managed account that hides API keys, is what separates it from OpenClaw and Hermes Agent, which are built around chat apps or a terminal and are usually run on your own model keys. The convenience has a cost: the easy path runs inference and integrations through TinyHumans' backend, so 'local-first' describes where memory is stored more than where your data is processed, unless you switch to local models and Privacy Mode and accept fewer features. It is also early beta with a very fast release pace, so it suits people comfortable with change more than teams that need stability.

About OpenHuman

OpenHuman, built by TinyHumans, is a Tauri desktop app around a Rust core. After sign-in you connect services such as Gmail, Slack, GitHub, Notion, or Google Calendar through one-click OAuth (118+ services, brokered through Composio), and every 20 minutes OpenHuman pulls fresh data from supported connections (those with native ingest, such as Gmail) into a Memory Tree: Markdown chunks, scored and folded into per-source, per-topic, and per-day summaries, stored in SQLite on your machine and mirrored as an Obsidian vault you can read and edit. That lets you ask things like "what's waiting on me?" or "what did Sarah say about the project across email and chat?" without pasting context in. On top of memory sit an orchestrator that spawns sub-agents with checkpointed, resumable runs, agent-proposed workflows that you review on a visual canvas before saving, a built-in toolbelt (web search, scraper, coder tools, browser and computer control, cron), voice dictation and spoken replies, and a Google Meet agent that transcribes calls into memory and can speak in them. Side-effecting actions such as sending an email or running a shell command go through an approval gate by default. The difference from OpenClaw and Hermes Agent is mostly about where your context comes from and who supplies the services: those are built around chat apps or a terminal and are usually set up with your own model keys, while OpenHuman is a GUI-first desktop app that builds memory from your connected apps and, by default, routes models, search, voice, and integrations through a TinyHumans account, with bring-your-own-key, local models (Ollama, LM Studio, MLX), and an enforced local-only Privacy Mode as options. The same core also ships as a terminal client and an embeddable Rust library. The tradeoffs: it is labelled early beta, the easy path depends on TinyHumans' backend, voice still leaves the device even in local-only mode, and the project is GPL-3.0 rather than MIT.