By Johannes Gudmundsson, CEO, inecta
Key Takeaways:
Not all AI is the same. Stochastic (generative) AI and deterministic AI serve fundamentally different purposes, and confusing them leads to costly mistakes.
ERP systems require deterministic AI. Generative AI belongs in the human layer around ERP, not at its core.
Rather than chasing AI hype, food manufacturers should be investing in clean, structured data and applying AI where it’s proven to deliver results.
The AI conversation is missing half the picture
Artificial intelligence (AI) is having a moment.
It’s everywhere: headlines, boardrooms, software demos. For leaders in food manufacturing, the question isn’t whether AI matters. It’s what this actually means for their business, and more specifically, what it means for Enterprise Resource Planning (ERP).
That’s where the conversation starts to go off track. Because when most people talk about AI today, they’re really talking about one thing: large language models. ChatGPT. Generative AI. Systems that can write, summarize, and respond in ways that feel remarkably human. But that’s only part of the story and treating it like the whole story leads to the wrong conclusions.
Stochastic vs. deterministic: Why the distinction matters
At a fundamental level, AI falls into two very different categories: stochastic and deterministic. Stochastic AI is probabilistic. It produces answers based on likelihood, not certainty. In many cases, there isn’t a single correct response, just a range of acceptable ones. That’s what makes it so effective for language, communication, and general productivity.
It’s also what makes it a poor fit for the core of ERP systems. ERP, especially in food manufacturing, does not operate in a world of probabilities. It operates in a world of exactness. There is one correct inventory number, one correct cost, and one traceable path for a product through production. These systems sit at the center of compliance, financial accuracy, and operational control. “Close enough” isn’t acceptable. If you input A, you need to get B every time.
That’s why ERP systems are, and will remain, fundamentally deterministic. This doesn’t mean generative AI has no place. It just means its role is different. Stochastic AI is incredibly valuable in the human layer surrounding ERP, the moments when people are interpreting information, asking questions, or trying to make sense of what the system is telling them. It can accelerate workflows, surface insights, and reduce friction in day-to-day tasks. But it improves the user; it doesn’t replace the system.
The real transformation is happening elsewhere.
Where the real transformation is happening
Deterministic AI, built on structured data, rules, and repeatable logic, is what will actually reshape ERP systems themselves. This is where AI moves from being a tool to becoming part of the infrastructure. Systems begin to do more than record what happened. They start guiding what should happen next. Production planning becomes more precise. Forecasting becomes more reliable. Cost tracking becomes continuous and granular.
ERP evolves from a system of record into a system of decision.
Good AI requires better data
None of this works, however, without the right data. And not just more data, but better data. For AI to function within an ERP, data has to be structured correctly, captured in detail, and readily available. If information is aggregated too early or lacks fidelity, the effectiveness of AI drops off dramatically.
You can see this clearly in costing. If costs are tracked only at a high level, the system loses visibility. But when costs are captured throughout the lifecycle of a product, from raw inputs through production to finished goods, the system becomes far more powerful. AI can only learn from what it can see, and it can only be precise if the data is.
This is also why infrastructure matters. Cloud-based systems aren’t just a convenience; they enable data to move, be processed, and be analyzed continuously. That flow enables AI to move from static reporting to real-time decision support.
AI won’t (and shouldn’t) replace human judgment
Even then, the role of people doesn’t disappear. AI can reduce errors and speed up decisions, but it doesn’t eliminate risk. In an industry where accuracy matters as much as it does in food manufacturing, there will always need to be human monitoring. There will always be someone validating outputs and making sure the system is doing what it should. The goal isn’t full automation. It’s better decisions, made faster, with greater confidence.
So, is AI ready for ERP? The answer depends on which AI you mean. Generative AI isn’t built to run ERP systems, and it doesn’t need to be. But deterministic AI is already moving into the core of these systems, and its impact will be significant.
The food manufacturing companies that benefit won’t be the ones chasing hype. They’ll be the ones who understand the difference, invest in the right data foundation, and apply AI deliberately, in the places where it actually works.
The CEO and founder of inecta, Johannes holds dual Master’s degrees in Mathematics and Computer Science. He has spent over two decades building ERP technology for the food industry. Johannes combines strategic acumen with a passion for endurance sports and chess, bringing the same long-game thinking to every product decision.
