Share with your CIO
East Africa’s AI adoption wave is real, but the region’s enterprises are at risk of building on sand. Writing for Business Daily Africa, NTT DATA East Africa’s head of applications argues that banks, factories, and governments rushing to deploy AI are skipping the foundational work that makes it pay off. The argument organizes around five layers: modernized core systems, clean and connected data, AI embedded in actual workflows, governance frameworks, and workforce capability built before the models arrive.
What this means for your business
The organizations most exposed to this argument aren’t the AI laggards. They’re the ones that have already signed the AI vendor contracts but haven’t asked whether their ERP data is clean enough to feed a model, or whether their procurement and finance workflows are documented well enough to automate. If your AI pilots keep stalling in production, the bottleneck is almost certainly upstream of the model itself, sitting in data quality or system fragmentation that no prompt engineering will fix.
The “embed before you expand” logic here is sound, and it holds regardless of geography. The recurring failure mode in enterprise AI looks like this: a proof of concept runs on a curated dataset, impresses the board, gets approved for rollout, then collides with the actual state of the organization’s data infrastructure. NTT DATA, as an implementation partner with incentives to sell modernization services alongside AI, naturally emphasizes the infrastructure prerequisite over the model selection question, but that tilt doesn’t make the underlying claim wrong. The sequencing genuinely matters. Gartner’s own research consistently shows data quality as the top barrier to AI value realization, not model sophistication.
The detail worth sitting with is the workforce framing. The piece positions AI literacy and change management as the final layer, but in practice they’re the one most often cut when budgets tighten mid-program. If your organization is three quarters into an AI transformation roadmap and people investment is still described as “upcoming,” that’s the leading indicator of a deployment that will technically work and operationally stall. The budget line to defend isn’t the model license. It’s the training program the CFO flagged as discretionary.
Based on reporting from AI alone won’t transform business but intelligent enterprises will, originally published 2026-08-09 14:00:00.
