Project Overview
| Client Industry | Manufacturing (Custom Metal Fabrication) |
| Business Type | Mid-size contract manufacturer, 3 production facilities, ~220 employees |
| Project Duration | 22 weeks |
| AI Service Provided | Enterprise Software Development |
| Technologies Used | React, TypeScript, .NET Core, PostgreSQL, SQL Server (legacy source), Microsoft Azure, Power BI embedded, REST APIs |
The Client Challenge
The client, a contract metal fabrication manufacturer running three facilities, managed production planning, inventory, and order tracking through a patchwork of 14 different spreadsheets, a decade-old inventory database, and a lot of institutional knowledge held by long-tenured staff.
The operational strain:
- Production schedulers manually cross-referenced spreadsheets across all three facilities to determine machine availability and material stock, a process taking 3–4 hours every morning before scheduling could even begin.
- Inventory data was updated manually and inconsistently across facilities, leading to situations where a job was scheduled against material that had actually been consumed by a different job the day before.
- Order status visibility for the sales team was poor answering a customer’s “where’s my order” question often required calling a plant manager directly, since no system gave sales real-time production status.
- Each facility had developed its own spreadsheet conventions over the years, making it nearly impossible to get an accurate company-wide view of capacity, work-in-progress, or on-time delivery performance.
- The company was bidding on larger contracts that required demonstrable production management maturity, and the current spreadsheet-based approach was becoming a competitive liability in RFP responses.
- Two key employees who maintained the most critical spreadsheets were nearing retirement, creating real risk of losing irreplaceable institutional knowledge embedded in undocumented spreadsheet logic.
Our Solution
Air Brite Labs designed and built a custom enterprise production planning and inventory management system that unified all three facilities onto a single platform replacing the spreadsheet patchwork with real-time, shared data and role-appropriate views for schedulers, plant managers, and sales.
Key capabilities delivered:
- Unified production scheduling — schedulers across all three facilities work from one shared view of machine availability, job queues, and material stock, eliminating the manual cross-referencing that used to take hours each morning.
- Real-time inventory tracking — material consumption is recorded at the point of use on the shop floor, keeping inventory data accurate across facilities instead of relying on end-of-day manual updates.
- Sales-facing order status portal — the sales team can check real-time production status for any order without calling a plant manager, closing the visibility gap that had been a recurring customer service friction point.
- Company-wide capacity and performance dashboards — leadership gets a real-time, accurate view of capacity utilization and on-time delivery performance across all three facilities for the first time, built with embedded Power BI reporting.
- Codified scheduling logic — the informal scheduling rules and priorities that lived in senior staff’s heads and spreadsheet formulas were documented and built into the system’s scheduling logic, reducing the retirement knowledge-loss risk.
- Role-based facility views — each facility retains views suited to its specific equipment and workflow, while feeding into the same unified data model company-wide.
Technical Approach
Architecture: Built as a .NET Core backend with a React and TypeScript frontend, deployed on Microsoft Azure to align with the client’s existing Microsoft-based IT environment and internal Active Directory setup.
Data Migration: Migrated historical inventory and order data from the legacy SQL Server database and multiple spreadsheet sources into a unified PostgreSQL schema, with a data validation phase to reconcile discrepancies between facilities before go-live.
Production Scheduling Engine: Built a scheduling module that codifies the client’s real prioritization rules (rush orders, machine changeover time, material lead times) extracted through direct interviews with schedulers into a system-supported scheduling workflow rather than a rigid, generic scheduling algorithm.
Shop Floor Data Capture: Implemented simple, tablet-based data entry stations on the shop floor for material consumption and job status updates, designed for quick use by machine operators without requiring extensive training.
Reporting: Power BI embedded directly into the application for capacity, inventory, and on-time delivery dashboards, giving leadership self-service reporting without needing to request custom reports from IT or engineering.
Integrations: Built REST APIs to connect with the client’s existing accounting software for order and invoicing data synchronization, avoiding duplicate data entry between the new system and financial systems.
Security & Access Control: Integrated with the client’s existing Active Directory for authentication, with role-based permissions distinguishing schedulers, plant managers, sales staff, and leadership views.
Implementation Process
- Discovery & Requirement Analysis (Weeks 1–4): Spent significant time on-site at all three facilities, interviewing schedulers and plant managers to document the real (often undocumented) scheduling logic and facility-specific workflow differences.
- Prototype / PoC (Weeks 5–7): Built a working prototype of the core scheduling module for one facility, validated directly with that facility’s scheduler against real historical scheduling scenarios.
- Development (Weeks 8–16): Built out the full system across all three facilities, including inventory tracking, the sales order status portal, and shop floor data capture stations.
- Integration (Weeks 17–18): Connected the accounting software integration and Power BI reporting, migrated historical data from the legacy systems.
- Testing (Weeks 19–20): Ran the new system in parallel with the existing spreadsheet process at one facility first, validating scheduling accuracy and inventory reconciliation before expanding.
- Deployment (Week 21): Phased rollout across all three facilities, with on-site training for schedulers and shop floor staff at each location.
- Optimization (Week 22 and ongoing): Refined the scheduling engine’s rule weighting based on real usage feedback and expanded dashboard views based on leadership requests.
Key Features Delivered
- Unified, cross-facility production scheduling with real-time machine and material visibility
- Real-time shop floor inventory tracking via tablet-based data capture
- Sales-facing order status portal eliminating manual status inquiry calls
- Company-wide capacity and on-time delivery dashboards via embedded Power BI
- Codified scheduling logic capturing institutional scheduling knowledge
- Role-based views tailored to schedulers, plant managers, sales, and leadership
- Integration with existing accounting software for order and invoice synchronization
Business Results
- Morning production scheduling time dropped from 3–4 hours to under 30 minutes across all three facilities
- Inventory discrepancy incidents (scheduling against already-consumed material) dropped by an estimated 90%
- Sales team order status inquiries to plant managers decreased significantly, freeing plant manager time for production oversight
- On-time delivery performance improved measurably in the months following rollout, attributed to more accurate scheduling and inventory data
- The company successfully won a larger contract citing its improved production management capability during the RFP process, directly referencing the new system
- Institutional scheduling knowledge was preserved and documented ahead of two key employees’ planned retirements, removing a significant operational risk
Technology Stack
| Layer | Technology |
|---|---|
| Frontend | React, TypeScript |
| Backend | .NET Core |
| Database | PostgreSQL |
| Cloud Infrastructure | Microsoft Azure |
| Reporting | Power BI (embedded) |
| Authentication | Active Directory integration |
| Integrations | REST APIs (accounting system) |
Why the Solution Worked
The project succeeded because it started with real time on the shop floor, not just requirements documents. Understanding the actual, often unwritten scheduling logic that experienced staff carried in their heads was the difference between building a generic production scheduling tool and building one that matched how this specific manufacturer actually operated.
Rolling the system out facility by facility, running it in parallel with the existing spreadsheet process before fully cutting over, also mattered — it let the team catch data and workflow discrepancies early without risking a disruptive company-wide failure on day one.
Future Scalability
The platform is designed to support the client’s continued operational growth, including:
- Adding predictive maintenance scheduling for production equipment using the now-centralized machine usage data
- Extending the sales order portal with customer-facing self-service status checking, not just internal sales team visibility
- Adding a fourth facility to the platform as the company’s expansion plans progress, using the same unified architecture
- Building more advanced capacity forecasting using historical scheduling and demand data now captured in one place
Because the system was built with a unified data model and modular scheduling logic from the start, extending to new facilities or adding new capabilities builds on the existing platform rather than requiring a rebuild.
Final Outcome
What was once a fragile, spreadsheet-dependent operation spread across three disconnected facilities is now a unified, real-time production management system that scales with the business and preserves institutional knowledge that was at real risk of being lost. The company now competes for larger contracts with the operational maturity to back up its bids.
