Product Deep Dive

From Intake to Insight: Building the SimplyChat Healthcare Report Engine

February 28, 2020 Simplyturn Editorial 4 min read
Report UI
Prototype, 2020

We built SimplyChat to convert messy EHR exports into a clean, shareable report a clinician can read in minutes.

Goal
Upload EHR → generate summary → share as a secure link.
Scope
Minimal AI/NLP. Deterministic rules + heuristics first.
Outcome
Clinician-ready PDF and web report with provenance trails.

The problem

Patients move across providers. Data scatters across portals and PDFs. Clinicians get duplicate meds, missing labs, and no timeline.

Architecture in 2020

Ingest (ZIP/PDF/CCD) → Parse (CCD/C-CDA, CSV, PDFs) → Normalize (FHIR R4 subset) → Aggregate → Summarize (rules + templates) → Render (web + PDF) → Share (time-bound link)

Privacy by design

  • PHI minimization on render; masked-FHIR bundle by default.
  • Short-lived, read-only share links with audit receipts.
  • On‑prem/VPC deployment. No third-party data processors for PHI.

Minimal AI in 2020

Rule-based extraction, ICD‑10/CPT tables, RxNorm mapping, light NLP for section detection. No heavy deep learning in the loop.

Team

Dedeepya
Dedeepya Sai Gondi — Co‑Founder/CTO. Architecture, privacy, and rendering.
Vamsi
Vamsi Krishna Reddy — Software Engineer. Parsing pipeline and FHIR normalization. Led a 7‑member dev squad.

What we shipped

  • Upload for CCD/PDF/ZIP
  • Automated report with clinical timeline and decision-support notes
  • Shareable expiring links with provenance
Security
  • TLS 1.2+, HSTS
  • RBAC + SSO
  • Audit receipts