Inspiration

Pharmacies handle medicines that have different batch numbers, expiry dates, suppliers, and dispensing histories. We identified a practical safety gap: when a batch approaches expiry or is recalled, it can be difficult to quickly determine the affected stock and identify customers who previously received that specific batch.

This inspired us to build PharmaFlow – Pharmacy Safety & Traceability, a system designed to connect inventory, batch tracking, dispensing audits, expiry monitoring, recall management, and patient safety workflows in one platform.

What We Built

PharmaFlow provides an end-to-end pharmacy management and traceability workflow:

Medicine Identification → Batch Registration → Inventory → Dispensing → Audit → Expiry/Recall Detection → Customer Traceability → Safety Action

Key features include:

Barcode/QR-based medicine identification
External medicine/product lookup when a barcode is not available in the pharmacy inventory
Pharmacist verification before adding externally identified medicines
Batch-level inventory and expiry tracking
Supplier invoice OCR and structured stock entry
Dispensing with batch-level traceability
Immutable dispensing audit records
Recall management and affected-customer identification
Expiry and low-stock monitoring
Patient safety communication workflows
Reports and pharmacy analytics
AI Pharmacy Assistant for operational pharmacy intelligence
What-If Expiry & Inventory Simulator for non-destructive decision support

The AI assistant is focused on pharmacy operations rather than diagnosis or treatment. It uses controlled pharmacy data to answer questions about inventory, batches, expiry, dispensing, recalls, suppliers, alerts, and operational trends.

How We Built It

The application was developed using:

React.js for the frontend
Node.js + Express.js for the backend
Firebase Admin SDK
Google Cloud Firestore for persistent pharmacy data
REST APIs for frontend-backend communication
Barcode/QR scanning for medicine identification
OCR for supplier invoice processing
Gemini-based AI for pharmacy intelligence
Transaction-based database operations for safe stock updates
Responsive UI for desktop, tablet, and mobile use

We designed the system so that important operations are handled by the backend and persisted in Firestore rather than relying on temporary frontend data.

For dispensing, the system maintains the relationship between the medicine, batch, quantity, customer, pharmacist, and dispensing time. This enables the system to trace which customers were associated with a particular batch when a recall or safety event occurs.

What-If Decision Support

One of the key features is the What-If Expiry & Inventory Simulator.

Instead of modifying live inventory, the simulator creates a separate projection using actual pharmacy data such as:

Current stock
Historical dispensing rate
Days remaining to expiry
Unit cost
Proposed additional stock
Demand changes
Supplier delays

It can calculate projected consumption, surplus, potential shortage, expiry risk, utilization, and capital at risk.

This allows pharmacists to explore scenarios without changing their actual inventory records.

What We Learned

During development, we learned that building a healthcare-related software system requires more than creating a functional interface. Data consistency, traceability, security, accurate status reporting, and reliable database operations are equally important.

We learned how to:

Design a full-stack application
Build REST APIs
Work with Firestore transactions
Implement batch-level data relationships
Integrate barcode scanning and OCR
Process real application data through controlled AI tools
Separate simulation data from live operational data
Handle authentication and pharmacy-level data isolation
Build responsive interfaces
Test interconnected modules
Design safety-focused workflows

Challenges

One of the major challenges was maintaining accurate relationships between batches, inventory, dispensing records, and customers. A recall cannot simply notify every customer; the system needs to identify customers who actually received the affected batch.

Another challenge was barcode identification. A barcode may identify a product but does not automatically provide pharmacy-specific information such as the physical batch, expiry date, quantity, or purchase price. PharmaFlow therefore separates external product identification from pharmacist verification and inventory registration.

We also faced challenges in ensuring that the AI assistant works with actual pharmacy data without becoming an unrestricted medical chatbot. We addressed this by using controlled backend operations and keeping the AI focused on operational pharmacy intelligence.

Outcome

PharmaFlow brings together pharmacy inventory management, batch-level traceability, dispensing audit, expiry intelligence, recall handling, customer exposure identification, safety communication, and AI-assisted operational decision support into one platform.

Our goal is to help pharmacies move from simply recording medicine transactions to maintaining a connected safety workflow where every batch can be tracked, every dispense can be audited, and potential risks can be identified earlier.