Case study: practical AI at workDiscuss your process
← All case studies

Birchwood · Therapy practice

From spreadsheet process to practice management portal.

How AI-assisted software development turned specialized business rules into a secure, role-based application.

SolutionBirchwood Management PortalBuild periodJuly 30–August 13, 2026PlatformNext.js application on Vercel

Birchwood’s important financial and operational processes lived across source-system exports, a complex Google Sheet, manual matching decisions, and separate provider communications.

In roughly two weeks, that operating model became a purpose-built portal for financial reporting, provider payments, documents, client notes, recurring tasks, access control, and more.

What AI did—and did not do

AI accelerated the design and delivery of the software. It does not make sensitive financial decisions inside the live application. Deterministic, reviewable code remains responsible for calculations, authorization, persistence, and payment-state changes.

The challenge

Preserve the practice’s exact rules.
Remove the recurring friction.

Different data shapes

Credit-card, cash/check, insurance, client-roster, and receivables exports used different schemas and naming conventions.

Specialized calculations

Provider obligations depended on an agreed percentage of revenue, subject to each provider’s individual monthly cap.

Controlled visibility

Providers needed timely self-service access to their own information—without ever receiving another provider’s data.

Administrative exceptions

Unmatched transactions, unpaid balances, documents, onboarding, and overdue work needed one consolidated view.

The solution

One operational system.
Two purpose-built experiences.

Administrator command center

See the whole practice and the work requiring attention.

  • Financial reporting and statement generation
  • Source-data review and name mappings
  • Payments, receivables, documents, and notes
  • Provider, prospect, intern, user, and task management
Private provider portal

Give every provider secure access to their own story.

  • Revenue trends and transaction detail
  • Amounts owed and payment history
  • Downloadable PDF statements
  • Balances, documents, notes, and assigned tasks

How it was built

Conversation became working software—in short, reviewable cycles.

A stakeholder described the process, reviewed real behavior, and supplied corrections. AI helped translate those conversations into specifications, code, tests, and documentation.

Initial provider-statement application and deployment workflow

Payments, role-based access, secure documents, provider metrics, and intern tracking

Client-grouped statements, shared documents, payment logic, earnings, navigation, and receivables

Client notes, provider prospects, unified management, payable statements, and payment instructions

Recurring tasks, financial controls, role dashboards, verification, and audit history

57repository commits≈9,500lines of TypeScript/TSX18automated test declarations

These figures describe implementation speed and breadth—not realized business outcomes.

Business impact enabled

Less reconstruction.
More operational clarity.

01

One source of operational truth

Financial activity, balances, statements, documents, and tasks come together in one portal.

02

Earlier exception detection

Unmatched payments and unverified submissions are made visible for deliberate review.

03

Provider transparency

Providers can inspect their own earnings, transactions, obligations, and payment history on demand.

04

Stronger process continuity

Formulas and workflows are encoded, documented, and less dependent on individual memory.

Measuring what matters

Capability is live. ROI measurement comes next.

The repository does not contain time studies, labor costs, adoption statistics, or before-and-after error rates. Birchwood can establish a baseline around these practical measures:

Statement preparation timeReconciliation effortCalculation correctionsProvider status requestsPayment cycle timeTask completionReceivables visibilityProvider adoption

Responsible use

Speed changes the build. It does not remove governance.

Human review remains essential for financial terms, mappings, payment verification, user access, and deployment decisions. The next phase focuses on measurement, security and storage hardening, automated build and backup controls, and observation across full monthly financial cycles.

That distinction matters: a rapidly delivered working system should have a clear path from successful prototype to durable business infrastructure.

“The value does not depend on putting a chatbot in front of employees. It comes from using AI to make tailored software economically and operationally attainable.”

Birchwood’s experience demonstrates a practical form of AI transformation for a small service business: turn institutional knowledge into a system, preserve human judgment, and create a stronger operating foundation.

What could we build around your business?