Teams that deploy AI save the equivalent of five hours per person per week, yet nine in ten executives report no measurable productivity impact at their own firm. Between those two findings sits the whole story of this month's articles: the gap between what AI demonstrably does and what organisations actually capture from it. July's three pieces work through that gap in sequence. Not Everything Needs AI starts before the tool decision, with the three questions I was asked about at the IoD Chartered Director Conference: what are you trying to do, how is the work done today and does it still need doing, and only then what tool fits. A well-judged no is what makes every yes credible. Governing the Redeployment Dividend picks up on the yes side and asks what happens to the hours AI recovers, because the evidence shows they leak: capacity is freed almost everywhere and value is recorded almost nowhere, unless the Board owns the redirection. The Balancing Item completes the account with the cost column nobody prices, the reviewing, correcting, and supervising that AI outputs demand, absorbed silently by people until it shows up as a distinct mental fatigue the research can now measure. Read together, they describe a single discipline: an honest ledger. What deserves remaking, what the freed capacity produced, and what the oversight genuinely costs. Boards that can answer all three are governing AI; Boards that can answer only the hours-saved line are celebrating a number that the other two columns quietly consume. If your time is limited, I particularly recommend Governing the Redeployment Dividend, because it names the accountability gap that explains the other two: AI owns execution, managers own the workflow, and unless the Board owns whether freed capacity creates value, the dividend refills with the work that was already there. How is your organisation accounting for both sides of its AI ledger, and who owns the answer to where the recovered hours actually went? - Mario | 
