DentalXChange

Engineering where being wrong has consequences

TYPESCRIPT
FASTIFY
PLAYWRIGHT
REACT
FASTAPI
CLAUDE

Summary

DentalXChange runs revenue-cycle management for dental practices and payers at national scale. LucidCo provides senior engineering across its credentialing and eligibility platforms — evidence-driven systems where correctness is the product.

A · Problem

Credentialing is evidence-driven paperwork: licences, federal database checks, insurer forms. Legacy automation had aged out — deprecated SDKs, race conditions, no monitoring story.

B · Approach

Rebuild the evidence pipeline as typed, tested services: automated verification capture against federal credentialing databases with managed sessions and health monitoring; queued eligibility processing; and Tag.Kit — a from-scratch React and FastAPI tool that maps insurer-form PDFs with Claude so humans tag them in minutes, not afternoons.

C · Artifacts

tests on the verification rewrite: 88
tests across the tagging tool: 550+
backend line coverage: 85%
gzipped initial bundle: 14 KB

A ground-up TypeScript verification service architected for a sixteen-scraper roadmap; staff portals and queued eligibility services; and Tag.Kit, built from scratch — PyMuPDF widget detection, pikepdf export, and a Claude tagging boundary hardened so attacker-controllable PDF text can never hijack the model's instructions.

D · Outcome

The credentialing pipeline runs on maintained, monitored, test-proven services. Capability stated plainly — client metrics are theirs to publish.

Plantactic