Endava finds rigid consulting and contracting models hinder the sustainability of AI projects in the Middle East

Field analysis of generative AI deployments over the past two years shows that the vast majority of initiatives remain stuck in the pilot phase, struggling to scale into production. While mainstream debate often attributes this stall to technical or regulatory friction, such as poor data quality, talent shortages, or immature governance frameworks, a less examined factor lies in the procurement mechanisms, consulting models, and commercial contracts that organisations have relied on for decades across conventional digital transformation programmes.

Regional acceleration and conventional procurement models

This friction is particularly acute across the Middle East amid rapid, state-led adoption pushes. While the United Arab Emirates has directed federal entities to roll out AI-powered services within 90 days, and Saudi Arabia continues to channel major capital into sovereign infrastructure and national capabilities through initiatives such as HUMAIN, many enterprises still approach AI procurement as standard transformation programmes with fixed specifications, predictable delivery milestones, and predetermined end states.

Zain El Haq, head of MENA at digital transformation and consulting firm Endava, argues that AI fundamentally alters project management conventions. Models evolve rapidly post-deployment, regulations and data localisation mandates shift continuously, and frontline employee adoption uncovers operational dynamics that cannot be mapped in advance. AI shifts from a static system delivered once to an ongoing capability requiring iterative tuning, turning demands for complete upfront certainty into an impediment to delivery.

Re-engineering contracts and shared risk

This shift requires rethinking project financing, governance, and value measurement. Legacy advisory models were designed to eliminate uncertainty prior to execution and tie fees to milestone delivery, a structure that falters when the underlying technology evolves rapidly, as seen in Abu Dhabi's fast-moving Falcon model iterations. A practical alternative structures engagements into sequential phases where each tranche validates hypotheses, delivers measurable business value, and shapes subsequent steps, tying commercial incentives to outcomes rather than procedural delivery.

Sharing risk and accountability between the enterprise and the advisor shifts operational behaviour and decision-making, giving both sides an incentive to reprioritise, adjust course, and direct capital toward proven use cases. Advisory partnerships become flexible frameworks supporting continuous adaptation, rather than rigid scopes defined before frontline learning begins.