Help for clinic staff

A visit in four steps

1. Open Session. Add customer or pick one from the list — or leave blank so Start creates a Walk-in.
2. Obtain verbal consent, then tick the short consent box. Start stays off until consent is ticked. Full scripts are below.
3. Start → treat and talk → Pause anytime (then Resume) → End when finished. Language is detected automatically; you do not pick it first.
4. Edit the AU physio note (Subjective, Medical History, Objective, Assessment, Plan, Treatment Today, HEP/follow-up, Precautions/red-flags). Drafts aim to keep clinical content only (not personal chit-chat) and may note when history came from a companion rather than the patient. Copy or Download for your practice software. Approve & delete audio when you are finished.

What we tell patients

Obtain verbal consent before Start. Pick the script that matches Settings (speech provider and whether local audio retention is on). Tick the Session checkbox only after they agree. Full repo wording also lives in docs/privacy/consent-script.md.

Eka live / Auto → Eka (usual clinic path)

I'd like to use Podlekar to help write your physiotherapy notes. It records our conversation, creates a draft progress note for me to check, and can prepare a short summary for you. The recording is sent to Eka Care in India for speech-to-text — outside Australia. Our clinic has confirmed with Eka that processing for this arrangement is in India and that Eka will not use this recording to train AI models. Podlekar cannot independently verify Eka's account settings or region. The recording is deleted here after I approve the note. You can say no and we'll document as usual. Is that okay?

Shunya Labs (only when Settings ASR is shunya)

I'd like to use Podlekar to help write your physiotherapy notes. It records our conversation, creates a draft progress note for me to check, and can prepare a short summary for you. The recording is sent to Shunya Labs outside Australia for speech-to-text. Their public cloud regions are the United States, Europe, and India — not Australia. Podlekar cannot independently verify where this recording is processed or whether it is used to train models. The recording is deleted here after I approve the note. You can say no and we'll document as usual. Is that okay?

Use this only for an explicit Shunya bake-off. Auto still uses Eka when Eka credentials are saved. Do not say the recording stays in Australia.

OpenAI Whisper (no Eka)

I'd like to use Podlekar to help write your physiotherapy notes. It records our conversation, creates a draft progress note for me to check, and can prepare a short summary for you. The recording is sent to OpenAI outside Australia for speech-to-text. The recording is deleted here after I approve the note. You can say no and we'll document as usual. Is that okay?

Demo / mock only (no live speech)

I'd like to use Podlekar to help write your physiotherapy notes. It records our conversation, creates a draft progress note for me to check, and can prepare a short summary for you. The recording is deleted after I approve the note. You can say no and we'll document as usual. Is that okay?

Clinic-day retention (local audio kept after Approve)

If Settings → Retain local audio after Approve is on, replace the deletion sentence in the script above with:

The recording stays on this clinic computer after I approve so our quality team can check transcription and note drafts — it is not uploaded to Podlekar cloud, not used to train public AI models by Podlekar, and not shared in chat.

Session home shows only a short muted reminder when retention is on. Turn retention off in Settings when you do not need research copies. Shared multi-user hosts should usually leave retention off.

Privacy — speech vendors & operator attestation

Live sessions send audio to the speech provider configured in Settings for transcription. Podlekar drafts a note for you to check; it does not replace your clinical judgment.

• Eka Care: audio goes to Eka in India for speech-to-text (outside Australia). The clinic has confirmed with Eka that processing for this arrangement is in India and that training is off. Podlekar cannot independently verify Eka’s account settings or region (the app still calls api.eka.care). Do not tell patients the recording stays in Australia.
• Shunya Labs: only when ASR provider is shunya. Audio is sent outside Australia. Public cloud regions are the US, EU, and India — Australia is not listed. A successful Test ASR only means the API key can mint a token.
• OpenAI: audio is sent outside Australia for speech-to-text.
• Operator attestation (Settings only): before live patients on Eka, complete the training + India-processing checklist and tick the attestation in Settings. That is a local clinic record — not an API guarantee. Test ASR only checks login. Re-attest if you previously ticked under Australia-hosting wording.
• Default: Approve deletes audio on this machine. Retention ON keeps local audio under data/audio/ for product review (gitignored). Never upload recordings to chat or git.

Practical guidance: docs/public/privacy-and-consent.md. Research vs clinical consent: docs/privacy/research-consent.md.

If something looks wrong

• Start greyed out — tick consent (customer is optional; blank creates Walk-in).
• Transcript says demo / mock — live speech is not connected. Ask whoever manages Settings. Do not file that text as the patient’s words.
• Invented measurements — delete them. You own the record.
• Patient declines — do not record; document as usual.
• Eka training + India processing — Settings must show the checklist completed before live patients. Test connection does not mean training is off or region is verified.