Case study
Building a Patient Intake Platform with AI-Assisted Processing
Paper forms and phone tag were stretching patient onboarding to two weeks. A custom intake platform with AI-assisted document processing cut it to days.
- Client
- Juniper Health — Placeholder
- Industry
- Healthcare
- Services
- Custom Software Development, Agentic AI Workflows
- Published
- 3x
- Faster patient onboarding
- 45%
- Fewer support calls
- 99.9%
- Platform uptime since launch
The Challenge
Juniper Health (a placeholder name for a multi-clinic healthcare provider) onboarded new patients the way most clinics do: PDF forms over email, printed forms at the front desk, and staff re-typing everything into the practice management system. Average time from first contact to first appointment was fourteen days, and the front-desk team fielded constant "did you get my forms?" calls.
Off-the-shelf intake tools couldn't model Juniper's referral workflows or integrate with its practice management system, so forms kept falling back to paper.
The Solution
We built a custom intake platform in three phases over four months.
Phase one delivered the patient-facing flow: a mobile-friendly guided intake that saves progress, supports document and insurance-card uploads, and sends automatic reminders when something is missing. Phase two connected the platform to the practice management system through a synced integration layer, ending the re-typing. Phase three added an AI-assisted processing step, with a document pipeline that extracts and pre-fills data from uploaded referrals and insurance documents, flagging low-confidence fields for staff review instead of guessing.
Every AI-extracted field shows its source snippet alongside, so staff verify in seconds rather than transcribe in minutes. Nothing enters the medical record without human confirmation.
The Results
Onboarding time dropped from fourteen days to under five — a 3x improvement that held as volume grew. Support calls about form status fell 45% once patients could see their own progress. The platform has run at 99.9% uptime since launch, monitored around the clock.
Tech Stack
The platform is a Next.js and TypeScript application backed by PostgreSQL via Prisma, hosted on AWS with infrastructure as code. The document processing pipeline uses the Claude API for structured extraction with confidence scoring, wrapped in a human-review queue built into the staff dashboard.
What's Next
The roadmap adds online scheduling and automated eligibility checks. If a spreadsheet-and-email process is quietly capping how fast your organization can grow, we'd like to hear about it.
