AI Summaries
Optional feature. Generates clinical summaries from form data.
How It Works
- Patient submits form
- Staff views submission
- Staff clicks “Generate AI Summaries”
- System sends form data to LM Studio (localhost)
- AI generates 3 summaries in parallel
- Staff reviews and edits if needed
- Summaries saved to submission
Three Summary Types
Clinician Summary
Key clinical information for staff review.
Highlights:
- Diagnosis and severity
- Treatment history
- Failed medications
- Contraindications
- Clinical decision points
Patient Summary
Plain-language explanation for the patient.
Explains:
- What they reported
- What happens next
- Treatment plan overview
PA Summary
Medical necessity summary for insurance prior authorization.
Includes:
- Diagnosis with ICD codes
- Failed prior treatments
- Clinical severity scores
- Medical necessity justification
- Treatment recommendation
Setup Requirements
LM Studio must be running locally:
- Download LM Studio from lmstudio.ai
- Load a medical model (e.g., Llama-3)
- Start the local server on port 1234
- Enable CORS in server settings
Without LM Studio running, the “Generate AI Summaries” button will fail.
You Can Skip AI Summaries
AI summaries are completely optional.
You can use IntakePilot without ever generating summaries. Staff can review raw form data directly.
Privacy
AI runs 100% locally on your machine.
Form data never leaves your computer. Nothing is sent to external AI services.
All processing happens through LM Studio on localhost:1234.
Editing Summaries
After generation, staff can:
- Edit any summary text
- Regenerate if needed
- Save changes
Edited summaries are stored in the database.
Technical Details
Summaries are stored in the processing_results JSONB field:
{
"summaries": {
"clinician": "...",
"patient": "...",
"pa": "..."
}
}
API endpoint: PATCH /api/v1/tms/intake/:id