How AI Helped Me Build a Custom Inventory System to Replace Spreadsheets

As an embedded development and training company, we constantly handle numerous electronic components — STM32 chips, resistors, capacitors, sensors, displays, and development boards. Previously we managed inventory with Excel spreadsheets: record purchases as additions, record issues as subtractions. In theory this works if everyone maintains the sheet diligently.

In practice, problems quickly emerged. Teachers would take development boards or sensors and forget to log them, or log incorrect quantities. By month-end the spreadsheet no longer matched physical stock. Concurrent editing by multiple people made it impossible to trace who changed what, when, and why — debugging inventory discrepancies became guesswork.

No More Searching Through Excel for Component Management

With the new system, administrators add components by specifying name, category, description, and stock quantity. For example, an STM32F103 development board belongs to the "development board" category; a temperature-humidity sensor belongs to "sensors". Information can be updated or deleted anytime. A search function lets users find any component instantly by name or category, eliminating the need to scroll through hundreds of spreadsheet rows.

In/Out Operations in a Few Clicks

The core requirement was simplicity: employees should need no training. When 100 sensors are purchased, the user selects the component, enters quantity 100, and clicks "Inbound" — stock increases automatically. When a teacher takes 20 sensors for a project, the same steps with "Outbound" decreases stock by 20. No manual calculation required.

A "Recently Used" shortcut provides quick access to frequently used components, avoiding repeated searches. An undo/redo feature acts as a safety net: if a user accidentally enters 100 instead of 10 for an outbound operation, they can undo the last action, which restores both the stock quantity and the operation record. Redo is also available if the undo was a mistake.

Package Management: Outbound a Full Kit at Once

Embedded projects often require kits of multiple components. For a smart car project, the kit might include two motor driver modules, one control board, two sensors, and other parts. Manually processing each item is tedious.

The system introduces "Package Management": administrators pre-define kits by grouping components with their required quantities. When a kit is needed, the user selects the package and clicks outbound — the system deducts all component stocks in one step. This significantly speeds up preparation of course materials and project kits.

Every Operation Traceable

Each inbound/outbound action automatically generates a log entry containing component name, operation type, quantity, operator, and timestamp — a complete audit trail. If a development board goes missing, the log shows exactly who took it and when, eliminating the need to ask around.

Users can add remarks to any log entry, such as "used for course recording", "smart car project testing", or "student lab materials", enabling later traceability of component usage.

AI Programming Is Changing How Developers Work

Technically, this system is not complex — no advanced algorithms or AI models. It solves concrete daily problems. Previously, faced with such a need, my first thought was to find off-the-shelf software or estimate development effort (frontend, backend, database, UI tweaks). Now, with AI coding tools, I focus on requirements: what problems to solve, how users will operate, where errors might occur — then let AI assist with code generation.

AI output is not 100% reliable, especially for inventory deduction, batch package operations, and permission checks. Testing and verification remain essential. Yet AI dramatically lowers the barrier for building small tools. Needs that once seemed not worth the development effort can now be prototyped quickly.

The real value of AI programming is not enabling everyone to build sophisticated software, but giving ordinary people the chance to turn repetitive, error-prone, low-efficiency tasks into working tools. This inventory system — component management, stock adjustment, package outbound, operation logs, data statistics — addresses exactly the problems we face daily. We used to adapt our business to software; now AI lets us make software adapt to our business.