Included with VibeCheck · Free download for Mac & PC
The AI-staffed back office for a therapy practice — run as kanban workflows on your own machine. Nobody hosts your data. This office houses nothing.
Shaula runs the practice's back office — website, blog, FAQ, review replies, newsletters, paperwork drafts — as kanban workflows a person approves. Open source, with an honesty gate on every word she drafts. Runs fully local for free, or on your own Google Cloud — either way, we never see your data.
Shaula Desktop — the app. Download, open, and the first-run screen walks you through everything (it even installs the free local AI for you). No account, no terminal, no fees, no telemetry.
Mac: Apple Silicon (main button) or Intel, macOS 12+, signed and notarized · Windows 10/11 64-bit — the installer is unsigned today (a signing certificate is on the list), so SmartScreen shows an "unrecognized app" notice once. Verify your download first with the checksums on the release page; then More info → Run anyway · ~120–130 MB app + a one-time local-AI download on first run. All downloads →
| Mode | Runs on | Cost | Privacy posture |
|---|---|---|---|
| Local | Ollama, this machine | $0 | Offline. Nothing ever leaves the box. |
| Cloud brain | Gemini on your own Google Cloud Vertex project | Your Google billing, pay-per-use | Your project, your data terms, your Google BAA if you need one. |
Every output passes a banned-language gate: no invented statistics, no outcome promises, no fake credentials. The linter refuses to ship them — and re-checks at review.
Anything that would be published, posted, or sent stops on a triage column until a person approves it. A model cannot approve its own review.
Website launch, weekly blog, FAQ page, review replies, referral outreach, psychoed handouts, fee letters, client comms scripts, directory profiles, practice paperwork, research desk, and more.
No clinical decisions, no crisis handling — those belong to the clinician, always. Shipped workflows are no-PHI by construction.
The desktop app needs none of this — download, open, done. The list below is for the terminal install only: macOS or Linux · Python 3.11–3.13 · git · Ollama (local mode) or the gcloud CLI + your own project (cloud mode) · Node 20+ for the board UI · roughly 2 GB for the harness and 5 GB for the local model. The setup walkthrough checks all of this before it touches anything, and never runs a billable command for you.