Forever Foundry Book a scoping call

Blog: Modernisation

Does your AI plan need legacy system modernisation first?

Published by Scott Seaborn, Founder

A warehouse team pilots an AI tool that suggests replenishment moves. The suggestions are good. Someone still has to key each one into a stock system installed fifteen years ago.

That is where many AI plans stall: not in the model, but in the systems it has to work through. The question for leadership is not whether to modernise everything first. It is which piece of legacy system modernisation the AI value actually depends on.

What the MIT and Mecalux study found

On 17 November 2026, the MIT Center for Transportation and Logistics and Mecalux published a study of 1,967 supply chain and warehousing professionals in 21 countries, from organisations with at least 100 employees that run their own warehouses. Fieldwork ran from 14 August to 1 October 2026.

40% cited legacy-system integration as the top barrier to scaling generative AI, and 54% said better integration with warehouse management and ERP systems would have the biggest impact. While 87% of generative AI use cases showed at least moderate improvement, only 39% had significant, documented gains.

These are self-reported views from warehouse operations, not measured returns, and they do not show that replacing old systems causes AI success. They do show where practitioners feel the friction. Read the research.

The FOFO™ view

AI integration with legacy systems is a business question before it is a technical one. A full replacement can take years and absorb the budget meant for the improvement itself. Doing nothing leaves people as the integration layer, re-keying what the AI suggests.

Imagine a manufacturer with an ageing ERP and a plan to use AI for demand forecasting. The forecast only has value if it changes purchase orders and production schedules. If planners must export, adjust and re-enter every figure, the gain disappears in handling. The legacy system modernisation worth funding may be one reliable interface to the planning module, not a new ERP.

The people who know where the joins break are rarely in the steering group: planners, warehouse supervisors, finance teams and the IT staff who maintain the workarounds. Ask the people who do the work before you scope any ERP modernisation.

Often several AI proposals depend on the same fix. Rank that shared integration work as an opportunity in its own right, rather than paying for it inside each project.

A legacy system modernisation test before you fund AI

  • Which decision or task will the AI change, and in which system does that change have to land?
  • How does the information move today: an interface, an export or re-keying, and how many hours a week does that take?
  • Is the data the AI needs complete and current in the legacy system?
  • What is the smallest modernisation step that removes the manual handoff: an interface, a data fix, a module upgrade or a replacement?
  • Would that step serve more than one AI or digital transformation priority?
  • Who owns the system, and who owns the business result?

How to measure AI ROI after the change

Take a baseline before you start: handling time per transaction, re-keying errors, lead time from recommendation to action, and the share of AI suggestions that are actually carried out. Record documented results separately from forecasts. That is the gap the study exposed: most teams saw improvement, far fewer could prove it.

AI Compass™ gathers insight through AI-powered surveys, branded for your organisation and sent by you, so the workarounds people live with become evidence: an AI readiness assessment grounded in how work actually happens. Slid3rs™ helps your people rank opportunities, including shared modernisation work, by business value and delivery complexity. FlowRoom™ brings leadership together to agree priorities.

Treat legacy system modernisation as part of the AI business case: fund the part of the estate your next decision depends on, and let the evidence decide the rest.

Related reading

Next

Find out which legacy systems are holding back your AI plans.

Explore AI Compass™