By Eben van Toder and Kristi Berger, 10 August 2025
Executive Summary: The Night A System Was Born
We developed a system that combines a simple, intuitive, common-sense manual tracking framework with advanced AI integration to give meat plants complete management of meat and other raw materials, applicable to any protein. The result is full batch number integration, real-time deviation reporting, and absolute process control at a fraction of the cost of conventional systems.
Our approach ensures information is available exactly when it is needed and in a format that makes sense. Every deviation, whether yield, cooking loss, or quality anomaly, is identified and reported immediately, allowing rapid corrective action and preventing downstream risks.
The system begins with understanding the process, the people, and the flow of raw materials. On this foundation, AI handles the heavy lifting: processing large, complex data streams, performing critical analytical tasks, and instantly integrating the outputs into clear, actionable reports.
What started as a late-night exchange of ideas with Kristi became the blueprint for a fully integrated, AI-enhanced manufacturing and management framework. It unites process control, yield optimisation, industrial engineering, and real-time decision-making into a single operational brain—delivering transparency, accountability, and efficiency without expensive equipment or annual licence fees.
Process and Product Management
We’ve built a deceptively simple yet extraordinarily robust manual system that records every transaction across the company and delivers complete tracking of products, ingredients, processes, and yields. Its strength lies in both its simplicity and reliability: even last week we were able to retrieve every single action from any department of a client company over the past two years—covering meat receiving, carcass deboning, packing, processing, and dispatch.
The most remarkable part was that this system stayed operational during a period when we were running an expensive South African plant software system, purchased to perform exactly the same functions. That imported system failed to meet even the most basic requirements we bought it for. The manual system never failed.
Over time, we improved it every few months, embedding new layers of full traceability, tracking every gram of meat and every ingredient through every process until it left the factory as a final product.
The Next Leap: AI as the Factory’s Nervous System
What Kristi and I built tonight was the next layer, an AI layer operating above the existing system.
Each day, every department’s production data would be collected through an AI interface and populated by the system into pre-designed Excel formats. From there, the data flows automatically into a centralised management dashboard covering every corner of the plant:
- Meat receiving
- Carcass deboning
- All processing, for example, Bacon, ham, and sausage production
- Stock levels in freezers and chillers
- FIFO management and shelf-life control
- Dispatch
The dashboard delivers daily intelligence to management, but the real leap forward comes from real-time anomaly alerts. If yields, throughput, or output values drift beyond a defined tolerance, management is notified immediately, allowing instant intervention rather than post-mortem correction.
From Personal Disputes to Measurable Outcomes
One of the persistent challenges in plant management is the tendency for strategy to become personal. When managers argue over acceptable bone percentages, cooking yields, or shrinkage tolerances, discussions easily turn into disputes about who is right rather than whether the standard is being met. For a long time, I wrestled with how to break this cycle and keep the focus firmly on outcomes.
The breakthrough came from Kristi, whose insight reframed the issue: strategy only becomes personal when it lacks an objective anchor. By linking operational standards directly to production data, every target—whether bone percentage, cooking yield, or shrink loss, becomes transparent, measurable, and indisputable.
Her solution was elegant and robust, fitting seamlessly with our system, which requires no expensive equipment or annual fees and delivers data instantly. She highlighted two critical ways to make strategy actionable:
HR Performance Evaluation per Shift/Team
Each production block can be traced to the responsible team or employee via batch numbers and timestamps. This allows income and losses to be compared between shifts, creating an objective basis for training, bonus programmes, or personnel planning. By combining production data with working hours, management gains a clear view of whether specific hours, days, or teams operate more efficiently.
QC Batch Tracking
Every deviation, such as excessive processing loss or temperature anomalies, is tied directly to the date, time, and batch. This enables rapid root cause analysis. Automatic QC alerts ensure that tolerance violations immediately trigger notifications, preventing defective products from moving further along the line. Over time, quality statistics reveal whether errors are increasing or decreasing and pinpoint where in production they occur.
What was once a riddle became a solution: management is no longer about opinion or confrontation but about transparent standards, traceable outcomes, and data-driven accountability.
Industrial Engineering: Matching Process to People
This system enables true industrial engineering of the production floor. We map factory layouts, product mixes, and workforce capabilities, then model workflows accordingly.
If the workforce is largely unskilled villagers, the workflow is designed like a village economy, intuitive and self-reinforcing with clearly defined micro-tasks. If it is a high-throughput European-style plant, we replicate precision, standardisation, and high-speed line balance.
In both cases, the only success metric is output per unit of resource.
Optimising Yields Without Losing Strategic Control
In deboning and other yield-sensitive operations, the best decision for the day’s orders may not be the most profitable decision for the business. Too often, junior staff are left to make those trade-offs.
Our new system removes that decision from the floor and gives it to pre-designated senior managers, operating within a structured decision-making model.
We adapted Robert G. Cooper’s Stage-Gate System, originally designed for product development in the 1980s, to the entire meat manufacturing process.
The Stage-Gate Adaptation for Factory Operations
1. Discovery
Identify operational decisions that could swing profitability up or down by cut choice, product allocation, or processing route.
2. Scoping
Evaluate the technical and market feasibility of each option. What happens if we consistently prioritise revenue-maximising choices? What is the opportunity cost if we do not?
3. Building the Business Case
Use hard data to model the impact on yield, revenue, and margins, and align with strategic goals.
4. Development
Redesign the process, station, or decision workflow.
5. Testing and Validation
Run controlled trials to confirm yield gains, operational stability, and client satisfaction.
6. Launch
Fully implement the system and lock in performance metrics. Even when the decision is made for lower-profit outcomes such as strategic client satisfaction, management can see and price the opportunity cost.
Intelligent Resource Allocation
A core feature of the strategy is staff capability mapping. We evaluate every employee for strengths and align them to the department where they deliver maximum value. This transforms human resource deployment from reactive placement to performance engineering.
QC as a Live Metric
In the new system, quality control is not just about defect rejection. It is integrated into yield and throughput analysis. Parameters such as product look, feel, and functional performance are captured in the same data flow as output figures, enabling instant cause-and-effect visibility.
Breaking the Company into Its Molecular Structure
To design such a system, the company must first be deconstructed to its smallest operational units, just as amino acids form proteins.
Every department is mapped by inputs, processes, and outputs. Each process is further split into micro-processes. We study these micro-units individually before recombining them into optimised larger structures.
This molecular view of operations allows redesign at the smallest level before scaling improvements plant-wide.
AI-Enhanced Continuous Improvement Cycle
The Stage-Gate model is applied again, this time to process improvement itself.
- Discovery – Identify the smallest units and sub-units.
- Scoping – List possible alternatives, pros, cons, and projected gains.
- Business Case – Justify the most promising alternative and set measurable targets.
- Development – Build the improved system or sub-system.
- Testing and Validation – Trial it on a limited scale.
- Launch – Implement fully, measure, refine, and repeat.
Each cycle becomes faster and more precise with AI analysing historical data, simulating outcomes, and identifying improvement hotspots.
The Modular Nature of the Approach
The strategy itself is modular, allowing it to be implemented in phases rather than all at once.
We could, for example, begin with the manual system as the operational backbone. Integrated into this system would be AI-facilitated uploads, where data recorded in books is transferred into digital formats, populating management dashboards and triggering real-time deviation notifications.
Other elements, such as integrated quality control, staff capability mapping, and full line and departmental redesign and optimisation, could be introduced in subsequent stages. This staged approach allows the company to adopt the system at a pace aligned with its resources, change management capacity, and strategic priorities, while ensuring that each phase is stable before moving to the next.
Conclusion: Strategy as a Living System
That night’s discussion with Kristi was more than just another business conversation. It was the moment our entire operational philosophy crystallised.
What started in Lagos as a robust manual system has evolved into a multi-layered, AI-supervised industrial management framework, one that merges the human ability to see nuance with machine precision in detecting trends, anomalies, and opportunities.
The result is a living strategy, responsive, data-driven, and capable of reshaping itself as the company grows.