From Formulator to Architect: Lessons from AI for the Modern Meat Scientist

By Eben van Tonder and Christa van Tonder-Berger, 20 May 2026

Students and researchers working with beef carcasses in a meat science teaching and research facility at the University of Nebraska–Lincoln. The setting reflects the practical interface between raw material assessment, yield evaluation, and system level decision making in modern meat science. (https://animalscience.unl.edu⁠)

Table of Contents

Introduction

A recent report on engineering teams at Google describes a decisive shift in how work is performed. Artificial intelligence now produces a substantial portion of software code, and the value of the developer has moved away from writing syntax toward judgment, system design, and the management of complex interacting processes. The observation that judgment is becoming more important than programming captures a broader industrial transition [1].

This shift is not confined to software.

In our own work, Christa and I increasingly find ourselves stepping away from formulation in the narrow sense and into the design of systems. The challenge is to build structures that track yield accurately, stabilise production, and formalise standard operating procedures within a framework that can manage complexity without becoming fragile.

As a direct result of this shift, our discussions have changed. It has become an ongoing, almost continuous line of inquiry between us. We find ourselves repeatedly asking what elements best manage a complex system. What features must be built into such systems so that they remain stable under pressure, variability, and scale. These questions are no longer abstract. We actively work to embed the answers into everything we design, from formulation frameworks to processing lines and yield tracking systems.

The questions have changed. Instead of asking what to add, we ask what the system must deliver. Instead of correcting failures, we aim to design processes that do not fail under normal variability. Instead of refining recipes, we define architectures that align raw material, process, and people.

In this, the formulation meat scientist is moving along the same path as the modern engineer.

From Formulation to System Architecture

Traditional formulation centred on composition. Salt, phosphate, protein, and water were balanced to achieve binding and yield. This remains necessary, but it is no longer the defining skill.

The work now begins with structure.

At the centre lies an understanding of binding systems. Two dominant frameworks govern most processed meat products. The first is based on myofibrillar protein extraction, where myosin plays the central role. The second is based on hydrocolloid systems, where binding arises from external structuring agents.

These systems operate according to different principles and must be understood as distinct architectures.

Binding systems and their functional logic

Myofibrillar systems rely on the extraction and alignment of muscle proteins. Salt increases ionic strength, mechanical action exposes binding sites, and heat stabilises the resulting network.

Hydrocolloid systems depend on hydration, dispersion, and gelation mechanisms that are often triggered by heat or ionic conditions.

The practical consequence is that the eating experience is defined by the system selected. Bite, juiciness, cohesion, and mouthfeel are direct outcomes of the underlying binding architecture.

The formulation is therefore not a list of ingredients. It is the expression of a system.

Activators and Control Points

Each system depends on specific activators. These determine whether binding occurs effectively.

In protein based systems, salt initiates extraction, mechanical energy distributes and aligns proteins, and heat completes gel formation.

In hydrocolloid systems, water availability governs hydration, while temperature and ionic conditions trigger gel formation.

The critical task is to align these activators with the available process. A mismatch between system and process is one of the most common causes of failure in production environments.

Limits and Failure Points

All systems have boundaries.

Protein extraction is limited by raw material quality, connective tissue density, and temperature control during processing. Gel formation is limited by water distribution, competing interactions, and insufficient activation.

These limits define the performance ceiling.

No adjustment of ingredients can compensate for structural failure. Excess phosphate cannot correct inadequate extraction. Hydrocolloids cannot stabilise a system that lacks a coherent protein network.

The ability to recognise these limits early distinguishes effective system design from reactive troubleshooting.

Historical Perspective and the Changing Raw Material

The modern raw material differs fundamentally from that of previous generations. Changes in breeding, feeding, slaughter age, and handling have altered muscle structure and water holding capacity.

Conditions such as PSE meat and DFD meat are now widely encountered across global supply chains. These conditions introduce variability that standardised industrial processes often struggle to accommodate [2][3].

Historically, alternative approaches existed. Extended curing, different processing sequences, and the use of functional connective tissue fractions were part of the working knowledge of earlier practitioners.

Many of these methods fell out of use as raw material became more uniform in industrial systems. Yet current variability suggests that these approaches should be reconsidered within a modern scientific framework.

Reconsidering Processing Architecture

Our own work has increasingly focused on processing architecture rather than isolated formulation adjustments.

This involves stepping back and asking whether the problem should be solved at the level of ingredients or at the level of process design.

In many cases, improvements are achieved by:

redefining process sequences
stabilising mixing and cutting conditions
simplifying standard operating procedures
integrating yield tracking systems

Older concepts such as Salzstoß, where connective tissue fractions are pre treated and incorporated into the system, provide practical solutions when aligned with current understanding of collagen behaviour.

The relevance of such methods lies not in their historical origin, but in their functional fit within modern processing environments.

Lessons from Software Engineering

The example from Google illustrates a broader principle. When execution becomes automated, value shifts to decision making.

The modern engineer defines objectives, anticipates failure, and manages systems rather than performing every technical step manually [1].

The same applies to meat science.

The formulation scientist must move beyond recipe construction toward system definition. This includes aligning raw materials, machines, people, and formulations into a coherent structure capable of delivering consistent results.

Each component behaves as part of a larger system. The role of the scientist is to ensure that these components operate together within defined limits.

Conclusion

Meat science is entering a phase where formulation alone no longer defines expertise.

The central challenge is the management of complexity arising from biological variability, process limitations, and economic constraints. Addressing this requires a shift from ingredient focused thinking to system level design.

The formulation meat scientist becomes an architect. The work extends beyond composition into the design of processes that remain stable under real world conditions.

The lesson from modern engineering is direct. As execution becomes easier, judgment becomes more valuable.

Those who understand and design systems will define the next phase of the industry.