What OpenHuman is (and isn’t)
- · OpenHuman is software you install on your computer—not a generic website chatbox.
- · You connect apps (like Gmail or Slack) one at a time; each asks for its own permission screen.
- · After data syncs (~20 minutes per cycle early on), you can ask plain questions that use that context.
In plain English
- Download and open the TinyHumans desktop build from their download page ↗.
- Sign in once to the product, then add each app you want separately—similar to connecting Google to a trusted calendar app (login to OpenHuman ≠ automatic mail access).
- The app pulls updates on a rhythm (about twenty minutes between passes in official docs). Right after linking Gmail, patience matters more than tweaking AI settings.
- You ask everyday questions—“what did I miss?”—once that first ingest pass has happened; answers draw on the notes the app builds on your machine, not a mystery cloud dossier owned by this guide site.
Technical snapshot (expand if you audit systems or code)
Executive snapshot
UI-first desktop shell: authenticate with the product, approve OS prompts (Accessibility, microphones, cameras when using voice or meeting features), wire OAuth integrations (mail, Slack, GitHub, Notion, calendars, ticketing, Stripe, drives, Linear, …). Automated fetch (~20 minute cadence in docs) fills the Memory Tree (organized notes on disk, backed by SQLite), compresses payloads for model context—then you prompt against that material. Humans can tweak notes in Obsidian-compatible storage when they want finer control.
Installation vs outsourced rollout SaaS
Consultants sometimes sell onboarding as a billed workspace—that is separate from GPLv3 redistribution and from this handbook. Here we assume you installed from TinyHumans GitHub / Download portal ↗.
Capability anchors
- Interfaces & voice — mascot, push-to-talk, Meet agents; optional surfaces widen the permission checklist.
- Memory posture — local summaries, vault editing, ingestion loops refresh context.
- Toolbelt — coder utilities, scraping, integration catalog upstream.
- Models — hosted routing; optional local models (see upstream GitBook/Ollama notes).
- Transparency — GPLv3 codebase; propose changes through Git—not silent forks.
Deeper diagrams live in TinyHumans’ Documentation ↗.
Suggested next hops
- Unified download & install →Start-here
- Hardware / accounts →Requirements
- OS-specific footnotes →Windows·macOS·Linux
- Operational FAQ →FAQ