Practice Ledger: Healthcare Billing Automation
A HIPAA-conscious billing pipeline that moves a therapy practice's SimplePractice exports into QuickBooks without anyone retyping numbers, and without patient data ever reaching an AI model.
Billing automation for a group practice
Built for an insurance-billing therapy practice in the DC area that runs SimplePractice for scheduling and billing and QuickBooks for the books. The same pipeline is designed to be replicated for other practices on the same two systems.
What needed to be fixed.
Every billing cycle, staff downloaded CSV exports from SimplePractice, stripped patient details by hand, worked out per-clinician income allocation, and typed the totals into QuickBooks. Hours of careful, repetitive work, with a real risk of a copy-paste mistake and a standing worry about protected health information touching any outside tool.
How we tackled it.
We split the work into two streams. Aggregated financial reports that carry no patient data feed QuickBooks journal entries directly. Transaction-level exports go through a de-identification step first, so any analysis (denials, allocation, reconciliation) happens on records with names and dates of service removed. AI runs inside the practice's own AWS environment under a Business Associate Agreement, and monitoring alerts go to the practice by default.
The approach and structure we used.
Here's how we thought about the implementation, the choices we made, and how we delivered it, without sharing anything that would compromise client privacy.
Mapped every report the practice actually uses and classified each one as PHI-free or PHI-bearing before designing the flow.
Built a secure upload portal into S3 for the exports SimplePractice can't deliver by API, with file-type detection so the wrong report can't land in the wrong folder.
Wrote the de-identification and allocation logic in Python, with per-clinician income following payment date rather than service date to match how contractors are paid.
Posted the results to QuickBooks through its API and matched deposits from the practice's banks against SimplePractice records for reconciliation.
Tools and platforms we used.
What this created for the business.
- Zero records of protected health information sent to an AI model: everything the model sees has already been de-identified.
- No more manual retyping between SimplePractice and QuickBooks; staff upload a report and review the result.
- Per-clinician allocation and bank reconciliation, the practice's two biggest time sinks, run as part of the same pipeline.
- A repeatable design for other insurance-billing practices on SimplePractice and QuickBooks.
We share what matters, keeping the sensitive details private.
The practice's volumes and financial figures are confidential. The architecture (two streams, de-identification before analysis, AI inside the customer's AWS account under a BAA) is the part that transfers to any practice.
Keep reading.
Apex Steward: AI-Augmented IAM Governance
A multi-tenant IAM governance platform built for mid-market security teams, where AI ingestion turns vendor documentation and policy PDFs into a working role matrix, application inventory, phased rollout plan, and NIST CSF-aligned maturity dashboard, collapsing what used to be a multi-month consulting engagement into days.
Read the case studyApex Audit: AI-Powered Automation Diagnostic
A self-serve AI diagnostic that scores a business across 9 operational categories in roughly 4 minutes, ranks its top automation opportunities, and, at the Deep tier, generates a personalized implementation roadmap with ROI projections. Replaces a $5,000 to $25,000 consulting engagement with an instant scorecard and a $50 paid report.
Read the case studyReady when you are.
No pitch decks. No twelve-step sales funnel. A conversation about what you need and whether we're the right fit.