Most AI programmes start with an ambition and a vendor meeting. Few start with an honest inventory of what the organisation already runs, who owns it, what it costs and what it has actually changed. Thirty days is enough to establish that picture, provided the diagnostic looks at evidence rather than at opinions.
The evidence is of five kinds: the systems and the AI already embedded in them, often by vendors and often unknown to the data teams; the budget lines, including recharges that hide where the work really happens; the platform, what it can serve today and what it cannot; the incidents of the last months, which show where accountability is missing; and interviews, read against the four other sources rather than taken at face value.
The output is not a maturity score. A composite score flatters and hides. The output is a map of decision rights with the gaps named, a short list of use cases sorted by what can be proven, and a handful of gates the organisation can apply itself: is there a supported system of record, has the data been examined, is there a named owner with time, is the funding committed, is the first step cheap and reversible.
A diagnostic that ends with a slide deck has failed. It should end with owners, dates and the condition under which each recommendation is reversed — and with a budget request written so that no work starts without a named owner and a funded commitment.
Was this useful? One click, no account.
Want to take it further? Contact us about this · Share on LinkedIn