Understanding AI’s Next Frontier in Food and Beverage

As the food and beverage sector races to understand what artificial intelligence can offer, FaBA’s Joel Andrade works through some of the common traps of using AI to grow your business.

A PhD student in the FaBA Innovation Pathways Program supervised by Professor Janet McColl-Kennedy, Joel says most businesses are still treating automation and AI as the same thing. They are not: they’re complementary tools that work best together.

“Automation is rule based,” he says. “Someone manually codes every action into the system. AI is different. It learns patterns in your data so it can make predictions and even take actions without an engineer telling it what to do.”

It’s an important distinction, especially as many businesses assume that using a programmed workflow on the factory floor means they’re already using AI.

Joel says the real shift happens when businesses stop thinking about AI as a smarter version of their existing setup and start seeing it as a set of tools that can solve very specific bottlenecks.

He describes general purpose tools like ChatGPT or Copilot as “just the tip of the iceberg”. They work in natural language, which is why they’re so visible, but they represent only one corner of the AI landscape. Behind the scenes are models that work with images, audio, operational data and highly specialised datasets. Whilst LLMs (Large Language Models) serve as a great entry point into the world of AI, richer domain-specific AI is embedded directly into physical and operational processes.

To illustrate what this looks like in practice, Joel points to new meal vision technology. In large catering environments like hospitals or aged care facilities, malnutrition is a persistent problem, and manual solutions were unworkable because of labour shortages. An AI driven vision system was deployed that scans finished plates, links them to individual patients, identifies what they’ve eaten and predicts nutritional risks before they occur. “It’s something you could never achieve manually at scale,” Joel says.

He adds that this kind of AI is not DIY. “You can DIY to prototype and test through methods such as vibe coding (coding in natural language),” he says, “but for commercial deployment, reach out to the right expertise. The implications of getting it wrong can be severe.”

Part of businesses navigating the new technology from a customer experience (CX) perspective is understanding the four Cs of consumer-oriented AI: communication, collaboration, convergence and control. Joel says these principles help businesses understand how AI interacts with users, how it manages and combines different sources of information and how it ensures systems remain safe, explainable and human overseen.

Before any of that, though, he says there’s one critical starting point: data.

“Every conversation with industry comes back to the same thing. Their data is scattered – across spreadsheets, planning systems, warehouse logs, everywhere,” he says. “If a business wants to move from automation to AI, they need structured, consolidated data. Without it, even the best AI tools can’t create value.”

Digital twins are where AI moves from insight to action. Poised to deliver major benefits to the food and beverage sector, digital twins are dynamic virtual replicas of real‑world operations such as production lines or even customer experiences, that draw on real‑time data and simulations to predict issues before they occur and test improvements safely and quickly across multiple datasets.

Professor Janet McColl-Kennedy, FaBA Innovation Pathways Program Lead, is the lead author on ‘Digital Twins: A Game Changer in Customer Experience’ in the Journal of Service Management. The article explains that, unlike static dashboards, a customer experience digital twin continuously analyses live data streams to anticipate problems and evaluate solutions before they affect customers.

For food and beverage businesses, this means being able to trial process changes, identify pain points, forecast customer responses, offer alternative solutions to customers and identify operational risks without disrupting production.

“Digital twins shift businesses from reacting to problems to predicting and preventing them. Our roadmap (AAICE) offers organisations a clear way forward to design and deliver better customer experiences with far greater confidence and control,” says Professor McColl-Kennedy.

Her team’s AAICE roadmap offers a step-by-step guide:

• Assess where inefficiencies or customer frustrations occur
• Analyse the data you already have
• Identify where a digital twin could add the most value
• Co‑design models tailored to your customers and operations
• Evaluate results and refine over time.

The outcome is a more proactive, data‑driven way of improving products, services and experiences – something the food and beverage sector can increasingly benefit from as consumer expectations continue to rise.

Regardless of the technology you are using, Joel and Janet have an important piece of advice: don’t begin with the technology. Begin with the problem. And then bring in the right expertise, using the right technology, to help you solve it.