FAQ — installs, connectors, Beta reality

· Straight answers distilled from onboarding flows and common Discord threads—not legal advice.
· When answers conflict upstream, README + TinyHumans release notes beat this page.

Install & updates

Where should I begin?: Follow the consolidated Start-here playbook—it mirrors TinyHumans README installers, download links, and both install scripts without juggling disconnected pages.
Do I need the terminal?: Prefer official GUI installers referenced in the README—the project markets a UI-first path. Scripted curl / PowerShell installers exist for automation but require reviewing raw scripts beforehand.
How often should I re-check hashes or scripts?: Beta cadence ships frequently. Before reproducing snippets in runbooks diff the latest scripts versus your saved notes (2026-05-17 editorial snapshot here).

Why does macOS want Accessibility?: Desktop automation surfaces—including voice hotkeys and agent tooling—rely on those OS hooks.
Can I postpone camera or microphone approvals?: Yes until you activate Meet mascots / voice workflows. Leaving them denied simply blocks those subtrees until you revisit First launch.

I signed into OpenHuman—why isn’t Gmail visible yet?: Signing in authenticates the product shell. Gmail (and similar) connectors require separate OAuth approvals and first auto-fetch windows (~ twenty minutes) before summaries appear credible.
What does “thin answers” mean right after onboarding?: Typical operator error is impatience, not hallucinations. Confirm auto-fetch logs or wait another cycle via Integrations guide.

Does everything stay on my laptop?: Workflow memory is designed locally, yet hosted models still ingest derived context when tasks route upstream. Read Privacy (plain English) before promising legal teams airtight air-gapped behavior.
Can I mandate Ollama-only routing?: README documents optional local models; wiring policy requires reading the current routing knobs in TinyHumans documentation rather than guesses here.