Share with your CIO
Future Processing, a Polish software delivery firm, spent 15 months turning individual AI habits into shared team practice, and published the full methodology and measurement data rather than a marketing summary. The core finding is blunt: 81% of staff were already using AI when the programme started, yet team-level AI practice scored 34 out of 100. By the second survey, that score reached 52.7, daily AI use doubled, and a library of delivery methods grew from 6 to 18 proven approaches. No company-level ROI is claimed, by design.
What this means for your business
Most enterprise AI programmes are sitting exactly where Future Processing started, high adoption numbers that look great in a board deck but mask a team-practice score that would embarrass anyone who actually measured it. The gap between “81% of staff use AI” and “34 out of 100 on team practice” is not a communication failure or a training gap. It is what happens when individual productivity wins stay with individuals. If your AI reporting lives at the usage layer, you are measuring the wrong thing and your programme is further behind than your dashboard suggests.
The finding that carries the most weight here is r = 0.42 between team-level AI practice and perceived project success, with individual use failing to predict the same outcome once you control for it. That correlation, modest but directionally clear, is the statistical case for why governance design matters more than tool access. Future Processing’s answer, two-day tool approval, centrally paid licences, and a Compliance Assistant pre-clearing roughly 150 common use cases, is essentially a friction-removal strategy. Shadow AI, meaning the use of personal accounts and browser extensions on work that IT cannot see, grows precisely where the approved route is slower than the workaround. The CIO who treats governance as a gate rather than a highway is manufacturing their own shadow AI problem.
The absence of an ROI figure is the most credible thing in this report, written as it is by a vendor pitching AI transformation services, which gives the conservatism on financial claims a commercial cost and therefore more weight than it would carry from a neutral source. Future Processing says ROI will be published after the new delivery standard covers most projects for a full year. That is the right benchmark to hold them to, and it is the right model to copy: set a measurement date before you commit the number, not after you need to justify the spend. If your AI programme has been running more than 12 months without a planned ROI reading, the question is not whether the number exists, it is why no one has committed to producing it.
Concept deep-dive: AI maturity scoring
AI maturity scoring rates not whether people use AI, but how well the practices behind that use hold up at the team and organisational level, graded on a rubric from “exploratory” to “transformational.” Think of it like a credit score for operational readiness: usage tells you someone has a card, maturity tells you whether they can handle a mortgage. Future Processing scores four areas, people, practices, tooling, and governance, quarterly, which turns a one-time survey into a trend line the business can actually act on.
Based on reporting from Lessons learned from a real enterprise AI transformation, originally published 2026-11-05 12:14:00.
