SINDHU KASHYAP SENIOR CONTENT STRATEGIST,

MIDDLE EAST & AFRICA

Published at the end of July 2025, ServiceNow’ s Enterprise AI Maturity Index lands with a paradox: hype is climbing while readiness slips. Surveying almost 4,500 leaders across five pillars, the average score fell nine points year on year to 35; fewer than 1 % cleared 50, and the top score dropped to 57.9.

Yet two-thirds report margin gains from AI – an average lift of 11 % – and 82 % plan to spend more next year, with budgets up about 8.6 %. Agentic AI is still early: roughly a third are piloting or live with at least one use case and about 40 % are considering adoption in the coming 12 months.

The story behind the numbers is part capability, part culture.“ The will is there; the scaffolding is not,” says Jessica Constantinidis, Innovation Officer – EMEA at ServiceNow, citing gaps in baseline skills, shared guardrails and the cadence to move from demo to production.

Since mid-year, posture has shifted from prohibition to enablement.“ The stance has flipped from‘ don’ t use it’ to‘ use it – safely’,” she adds, as private model access with DLP, audit trails and role-based controls gives people a secure sandbox to learn. Her near-term playbook is deliberately everyday: treat AI as a company-wide language, teach problem framing and verification, and make time savings legible so gains do not evaporate.

Velocity magnifies those organisational gaps.“ The headline reason is the velocity of capability change,” argues Kurt Muehmel, Head of AI Strategy at Dataiku; a roadmap drawn in January can feel dated by June. But progress is not magic – it is method.

“ Enterprises don’ t just want a result – they want defensible reasoning,” he says, urging provenance, re-playable steps and evaluation harnesses so an answer can

be audited like a junior analyst’ s model. Muehmel recommends cross-functional AI studios, early guardrails and operator-style metrics – minutes saved, defects avoided, dollars recovered – to turn

prototypes into systems.

Caution matters most where autonomy meets impact.“ Just because something can be automated doesn’ t mean it should be,” warns Levent Ergin, Chief Strategist for Climate, Sustainability and AI at Informatica, noting how misfiring agents in critical systems can cause real economic harm. And Ergin is firm on foundations:“ Data is to AI what ketchup and mayonnaise are to fries – it’ s fundamental,” a reminder that harmonised terminology, consolidated metadata and master data management make agents predictable rather than brittle.

The index’ s‘ Pacesetters’ show what good looks like. Averaging 44 versus 35 overall, they are far likelier to hit outcome – better experiences, higher productivity and faster innovation – and their patterns are reusable, so the second and tenth use cases ship faster than the first. They also make savings legible, turning time back into capacity.

So, the near-term path is pragmatic and human. Treat AI as a shared language; start with low-risk, high-volume work; keep humans in the loop and let autonomy be earned by performance. Measure relentlessly and reuse the wins. Published in late July, ServiceNow’ s report is both caution and catalyst: progress arrives when data discipline, governance and everyday enablement meet urgency – and when leaders resist fireworks in favour of foundations. p

18 INTELLIGENTCIO MIDDLE EAST www. intelligentcio. com