- AI
- ASSESS
- A5 · AI USE-CASE DUE DILIGENCE™
- 3 WEEKS
Four AI use cases evaluated.One proceeded.AED 4M saved from the three that didn't.
A UAE enterprise brought four AI investment cases to a three-week Assess engagement. One recommendation was to proceed. Two came back with redesigned non-AI paths capturing most of the value at a fraction of the cost. One came back with a "no-go" recommendation on data quality grounds. The fixed-fee A5 engagement paid for itself many times over in the investments it prevented.
N=1 ILLUSTRATIVE COMPOSITE — representative pattern for a UAE enterprise operating at scale. Details drawn from patterns across NexITC engagements and market data. Not a specific client narrative.
THE SITUATION
The situation
The client is a UAE-headquartered enterprise operating across the UAE and broader GCC, with a diversified operations footprint and a mixed customer base spanning multiple industry verticals.
The board had approved an AI investment programme for the fiscal year — driven by competitive pressure from regional players expanding UAE presence, and a strategic thesis that "AI-first operations" would be the operating model within 24 months. The COO had shortlisted four candidate use cases, each with a preliminary business case from internal teams and vendor briefings. Aggregate proposed spend: approximately AED 5.5M in year-one CapEx, with implied recurring spend of AED 8-10M annually at steady state.
THE SPECIFIC QUESTION
The specific question
The clinic conversation surfaced the shape of the ask precisely. The board would sign off within eight weeks. The COO needed three deliverables before that sign-off: a use-case-by-use-case go/redesign/no-go recommendation with evidence, a revised aggregate budget reflecting the recommendations, and the reasoning trail for each recommendation — so the board could interrogate the logic rather than accept a verdict.
The Practice Lead's scoping response confirmed A5. The client accepted the scope structure explicitly, including the "recommend against" clause. That acceptance turned out to matter.
APPROACH AND TIMELINE
Approach and timeline
The engagement ran three weeks against the four candidate use cases in parallel, with a shared evaluation framework. Each use case was scored against six criteria: commercial value at stake, data readiness, determinism of the target outcome, alternative non-AI paths, operational readiness of the receiving function, and blast radius if the AI approach failed.
Week 1 gathered baseline evidence — internal business cases, vendor briefings, current-state operational data, and independent data samples. Week 2 ran the scored evaluation in parallel across all four. Week 3 synthesised recommendations, walked the reasoning trail with the COO and process owners, and prepared the board readout.
Practice Lead attendance ran across all three weeks. CEO attended the board readout at the COO's request, given the aggregate spend and the "recommend against" positioning.
WHAT WE RECOMMENDED AGAINST
What we recommended against
The recommendation the engagement was scoped to be willing to make.
Three of the four use cases came back with recommendations that departed from the client's original investment thesis.
The first use case — an AI-based operational optimization proposed at approximately AED 1.4M — was recommended for redesign. The client's existing operational systems already ran heuristic-based optimization capturing most of the theoretically-available improvement. An AI layer would add marginal value at significantly higher integration and operational complexity. Redesign: a two-week configuration sprint by the internal team, closing 60-70% of the gap the AI investment claimed.
The second use case — a customer-facing AI assistant proposed at approximately AED 1.1M — was also recommended for redesign. Sixty percent of inbound interactions were three recurring pattern types, each solvable by rule-based automation with high determinism. Redesign: rule-based automation for those 60% first, measured for six months, then revisit AI approach for the residual complexity.
The third use case — AI-driven demand forecasting proposed at approximately AED 1.6M — was recommended as a "no-go" at that scope. Data quality evidence made forecast accuracy improvements unlikely to exceed the existing analytical baseline. The prerequisite recommendation: a Data Trust Sprint before revisiting the AI approach.
Aggregate spend redirected away from the AI programme: approximately AED 4.1M in year-one CapEx.
OUTCOMES
Outcomes
The one use case that proceeded — a route optimization application — moved to a B1 Pilot Factory engagement in the following quarter. Deployed year-one cost of approximately AED 1.2M against a targeted operational value of AED 3-4M in year-one cost reduction, subject to steady-state measurement.
The first redesign was delivered by the client's internal team over two weeks, achieving approximately 65% of the value the original AI proposal targeted at less than 5% of the proposed cost. The second redesign — rule-based automation on the recurring interaction patterns — deflected approximately 42% of inbound volume in the first quarter of measurement.
The forecasting use case remained on hold pending data quality remediation. Six months post-engagement, the client had commissioned an A6 Data Trust Sprint to address the underlying gap.
WHAT COMES NEXT
What comes next
The client's follow-on engagement pattern illustrates the assessment-first commercial model working as intended: one A5 engagement surfaced two follow-on engagements (B1 for the proceeded case; A6 for the data quality prerequisite), while three-quarters of the originally-scoped AI investment programme was redirected to paths the client's internal teams could execute.
Adjacent SKUs: A5 · B1 Pilot Factory™ · A6 Data Trust Sprint™
Have four AI use cases on the table and a board decision looming?
Book a clinic. Practice Lead attends. Three-week Assess engagement produces the evidence trail your board can interrogate — including "no-go" recommendations where the evidence supports them.