Food and beverage manufacturers don’t lack enthusiasm for AI. They lack connected foundations.
In a Barclays survey of 100 food and beverage leaders, 90% said AI could improve decision-making speed, yet only 49% said AI is part of their operations or forecasting infrastructure. Respondents named three practical barriers: data quality, legacy systems and a lack of analytics capability. TraceGains’ 2026 survey of 423 food quality, safety and R&D professionals points the same way: 40% identified disconnected systems and data as the top barrier to moving faster and just 9% described their organizations as fully connected.
Across manufacturing, the result is a payoff gap. A Columbus survey of senior IT and technology decision-makers found 83.4% are very or extremely confident AI can solve their most pressing business challenges. Yet increased productivity was the most common outcome reported (40.5%), while revenue growth (7.6%) and reduced operating costs (5.7%) trailed far behind.
The compliance deadline that doubles as an AI foundation
For many food companies, the most important data project for the next two years is already on the calendar. The FDA’s Food Traceability Rule, known as FSMA 204, requires companies that handle foods on the Food Traceability List to capture key data elements at critical tracking events and produce records within 24 hours of an FDA request. The compliance date is now July 20, 2028, largely because companies’ readiness depended on trading partners with less advanced systems.
It is tempting to treat FSMA 204 as a compliance checkbox. The better move is to treat it as infrastructure. The same clean, connected item, lot and supplier data that answers an FDA request in 24 hours is the data that demand sensing, quality prediction and AI agents need to work.
Litehouse Foods is building that foundation now. Halfway through a major ERP deployment, the company is mapping processes to how work happens on the shop floor and has formed a data governance committee to bring consistency to item, supplier and product data. “Clean data is essential for future tools,” says Andy Juarez, vice president of operations.
Clean data alone won’t close the gap
Even with good data, many plants hit a second wall: the organization. AI can surface a supplier risk or a lot-level anomaly in minutes. If the response still runs through functional silos, sequential handoffs and layered approvals, much of that speed is lost before anyone acts. Automating work inside the same silos doesn’t eliminate slow decisions or organizational friction. Closing that gap requires more than connected systems. It requires an operating model that brings people, decisions, expertise and work together around the outcome.
That is the gap Columbus’ Rapid Adaptation Engine (RAE) is designed to address. RAE is not a software system. It is an organizational-design approach that complements technology by changing how people, decisions, expertise and work come together around critical customer and business outcomes. “Long-term value will come from redesigning work around the combined strengths of people, data and AI,” says Ed Kunnary, the US lead on Columbus’ Strategy and Growth team.
Five moves from traceability to AI value
1. Treat traceability data as shared infrastructure. Lot, item and supplier records should have business owners, not just QA or IT custodians.
2. Fix the process before you automate it. Map receiving, transformation and shipping as they happen on the floor, workarounds included.
3. Set the value target first. Tie each initiative to a measurable outcome, such as hours to trace a lot, waste avoided or on-time, in-full performance.
4. Define decisions before the alert fires. Agree in advance who can place a hold, switch a supplier or reallocate inventory when AI flags an issue.
5. Start where coordination matters most. Traceability, recall readiness and supplier quality are natural first areas for cross-functional teams.
Don’t wait for 2028
Do not wait for perfect readiness. Choose one traceability or quality challenge and turn it into a working AI agent through Columbus’ Idea to Agent program. For more on building the organizational foundation, download C|The Shift: Redefining Global Manufacturing.
Turn one traceability or quality challenge into a working AI agent through Columbus’ Idea to Agent program. Or download C|The Shift: Redefining Global Manufacturing for the full Litehouse Foods story.
Key takeaways
- Food and beverage leaders see AI’s potential, but disconnected data and legacy systems keep most use cases from scaling.
- FSMA 204 traceability data, due by July 20, 2028, can double as the foundation AI needs.
- Value depends on the organization too: clear decision rights and cross-functional teams turn AI insight into action.
- The companies that create value first will not be those that merely meet the traceability deadline. They will be the ones that use the work to connect data, clarify decisions and redesign how the organization responds.
