AI SaaS case study

VidalSigns

A HIPAA-aligned, white-label platform that turns a dense lab report into plain-English answers — and lets a clinic or a gym offer the same thing under its own brand.

  • IndustryHealthcare SaaS
  • UsersIndividuals and partner clinics
  • Data retentionPurged within 24 hours
  • ServicesProduct Design, Development, AI
VidalSigns: AI SaaS Health Platform

Overview

VidalSigns reads a lab report and explains it in plain English. Somebody uploads a PDF or a photograph of one, a multi-pass OCR engine pulls the values out of whatever template the lab happened to use, and a chat answers questions about what they mean. No clinical vocabulary required, and no account to create — access arrives as a secure link in an email.

The second half of the product is the part partners buy. A gym, clinic or wellness centre configures a branded microsite of its own — layout, theme, colours, hero copy, pricing — and hands its members a bundle of QR codes. The whole thing deploys from the partner dashboard without anybody at CodeMyPixel touching it.

Health data does not sit still here. Uploaded files and everything extracted from them are purged automatically within twenty-four hours, behind end-to-end encryption, role-based access and an audit trail — which is what makes the offer sellable to an organisation that has to answer for it.

  • React / Next.js
  • TypeScript
  • Tailwind CSS
  • PostgreSQL
  • Vector Search
  • Generative AI (LLM)
  • Advanced OCR
  • Stripe Billing
  • Vercel Hosting

Project video

vidalsignsClarity you can trust.

A walkthrough of VidalSigns, from an uploaded report through to a partner deploying a branded microsite of their own.

Key capabilities

  • Report ingestion

    PDFs and photographs both. A multi-pass OCR engine reads the values out, whatever the lab's template looks like.

  • Plain-English chat

    A retrieval-augmented model explains results against reputable medical sources, in a calibrated tone.

  • White-label microsite

    Partners pick a layout, edit every section, set their own pricing, and deploy the site themselves.

  • QR code bundles

    Access is distributed as printed codes and secure links, so a member never has to make an account.

  • Twenty-four hour purge

    Uploaded files and everything extracted from them are deleted automatically, on a timer nobody can extend.

The challenge

A lab report is written for a clinician, and then handed to somebody who is not one.

Individuals get a page of reference ranges and abbreviations with no way in, so the questions land on whoever is nearest — a gym's front desk, a clinic's phone line, a search engine at midnight. The organisations fielding those questions had no way to answer them at scale either: doing it properly means clinical safety, HIPAA-grade privacy and OCR that survives a phone photo of a printout, and none of that is a side project.

  • Jargon on every line

    Reference ranges and abbreviations that mean nothing without clinical training.

  • Every lab formats differently

    Templates vary, scans come in crooked and low-contrast, and extraction cannot assume a layout.

  • The same questions, endlessly

    Gyms and clinics answered the same explanations by hand, one member at a time.

  • Nowhere safe to put the data

    No partner could offer this without a privacy guarantee they were willing to put in writing.

Our solution

One assistant, two audiences: the person reading their own results, and the business offering that reading under its own name.

The individual side is a chat. Upload a report, get an explanation, ask follow-ups — reached from a link in an email, so there is no account to make and nothing to remember. The partner side is a dashboard: configure a branded microsite, set your own pricing, buy a batch of QR codes, and watch usage come back as analytics. Both sit on the same encrypted core, and both forget the file within a day.

RAG-backed answers

Explanations grounded in reputable medical sources, with a consistent safety disclaimer.

Multi-pass OCR

Image preprocessing plus table and field detection, tuned for the quality real scans arrive in.

Website builder

Three layouts, with theme, brand, hero, features and pricing all editable, deployed from the dashboard.

Secure link delivery

Results reach a person through an emailed link rather than an account they have to create.

Tenant isolation

Role-based access, strict separation between partners, and audit logs across the whole platform.

Impact & outcomes

From complexity to clarity.

A secure, scalable health-insights assistant that individuals can actually read, and that partner organisations can put their own name on.

98%
OCR accuracy on real report scans
<2s
average AI response time
100%
of uploaded data purged within 24 hours
40%
fewer questions reaching partner staff

What we had to solve

Diverse lab templates

Solution: A multi-pass OCR engine with image preprocessing and table and field detection, so extraction stays reliable across varied and low-quality report formats.

Safe medical explanations

Solution: A retrieval-augmented model aligned to reputable medical sources, combined with a calibrated tone and a consistent safety disclaimer.

HIPAA-grade privacy

Solution: End-to-end encryption, strict role-based access control, detailed audit logs, and an automated 24-hour file deletion policy.

Questions

The main goal was to create a dual-purpose platform: an AI assistant for individuals to understand their lab results, and a white-label SaaS solution for businesses to offer this service under their own brand, complete with custom settings and QR code distribution.

Ready to make your data make sense?

See how we can build your AI platform.