Building Systems for a Growing Clinic: A Three-Year Digital Transformation at a 110-Person Healthcare Organization

When I arrived at a 110-person private clinic in Budapest in summer 2023, the organization was running on digitally ad hoc solutions suited to its earlier, smaller scale. Fifty employees and sixty-plus subcontractor doctors relied largely on manual, paper-based processes. The clinic had grown to a scale where these manual approaches could no longer keep up with day-to-day operations.

The moment that made this clear came at the reception desk. When a doctor finished their shift, a receptionist would manually calculate their commission while patients waited in a growing queue. The doctor, anxious to leave, would press for the math to be done faster. It was inefficient, tense, and error-prone. But it was just one symptom of the fact that the clinic's growth had outpaced its digital infrastructure, visibility, and control.

Every solution had to be built within strict healthcare compliance requirements. Data governance, GDPR, and healthcare-specific regulations shaped every architectural decision. Patient data, staff records, financial information — all of it had to be stored, processed, and accessed according to regulatory standards. This wasn't a constraint to work around; it was foundational to why these solutions could be trusted.

Compliance wasn't the only constraint. This work unfolded against a difficult economic backdrop — a post-COVID environment of high inflation and a stagnating economy — which meant tight budgets and little room for expensive, off-the-shelf enterprise software. Every decision had to weigh cost as carefully as capability, favouring solutions that delivered real value without overspending.

I started with foundations. The first three months were spent migrating the entire organization from private Gmail accounts to Google Workspace, establishing proper access control, user management, and security groups. We tried Slack for internal communication, but it didn't fit this audience — Google Chat proved to be the right tool. This wasn't glamorous work, but it was essential. Nothing else could be built safely on top of uncontrolled identity and access — especially in a healthcare context where compliance depends on knowing who has access to what.

Once that foundation was in place, I turned to finance, which was perhaps the biggest pain point. The clinic received invoices from dozens of suppliers — doctors, delivery companies, material vendors — mostly on paper. Finance would collect these invoices and send them to the accountant. Regularly, invoices would get lost in transit or mishandled. The result: every year, the accountant sent back 600 to 700 invoices that had no digital images and therefore couldn't legally enter the books. It was a constant, time-consuming cycle.

The finance team was maintaining two handwritten spreadsheets where they manually entered every invoice detail. These spreadsheets dated back to the clinic's earlier, smaller-scale operations and hadn't kept pace with its growth — there was no data validation, no normalized date format, and no structure suited to downstream processing.

In mid-2024, I introduced an application that pulled invoice data automatically from Hungary's tax authority database. The clinic now knows exactly which invoices were submitted to them, with images already attached for many. For the rest, the system captures them from emails, manual uploads, or supplier follow-up — but this now represents a fraction of the effort it once did. The result: what once took two weeks to compile from scattered email history and was never 100 percent accurate is now compiled in real time, 100 percent accurate, and immediately processable by the accountant.

In parallel, I introduced an attendance management application that integrates directly with the accountant's payroll system. Employees can now see exactly what was recorded against what gets paid. Payroll errors that used to stay hidden now surface immediately.

In early 2025, I built a quarterly utilization dashboard using data engineering and Looker Studio. Every doctor now receives an automated email showing how their utilization numbers have changed quarter over quarter, whether their clinic availability shifted, all presented in a clean dashboard. Doctors got visibility they never had before.

But the real acceleration came later, when I started building custom web applications end-to-end with AI assistance. The material management tool uses OCR to recognize items by barcode and packaging, suggesting database matches so assistants don't have to type manually. They scan product packaging or delivery notes, the system recognizes what they received, and inventory updates automatically across two locations and specific rooms. Ordering moved from scattered email coordination to a structured, auditable process.

Cash management came next. Staff who received cash advances were losing track of spending — receipts disappeared, details vanished, money went unaccounted for. I built an application where users scan receipts with OCR. The receipt is captured immediately, the balance updates, and there's a complete audit trail. No more lost data.

Then the cashier desk reconciliation tool. The clinic has seven cashier desks across two locations. Receptionists need to capture every transaction in real time. Finance needs to reconcile cash, cards, and other payment types easily. Management needs visibility into actual income per desk. One application serves all three needs — data entry for receptionists, reconciliation views for finance, income reporting for management.

Three years. One cloud database. One data pipeline. One reporting layer in Looker Studio. Six custom applications built end-to-end. The result: a 110-person organization that went from manual, spreadsheet-driven operations to having real-time visibility, automated workflows, compliant data practices, and systems that actually work.