[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"article-body-everyone-has-ai-few-get-value":3},"\nEvery enterprise now owns AI. Copilots in the office suite, a chatbot on the\nintranet, a machine learning team somewhere in the data function, a slide in\nthe annual report. The adoption question is settled.\n\nThe value question is not. The MIT NANDA initiative's report, The GenAI\nDivide: State of AI in Business 2025, put a number on it that should be\nuncomfortable: about 95% of enterprise generative AI pilots deliver no\nmeasurable impact on profit and loss. S&P Global Market Intelligence found\nthat 42% of companies abandoned most of their AI initiatives during 2025, up\nfrom 17% a year earlier. Spending rose. Results did not.\n\n*The constraint was never intelligence. It was the distance between the\nmodel's output and a decision someone actually commits to.*\n\n## Where the value leaks out\n\nWatch what happens to a typical AI output inside a large company. A model\nproduces a forecast, a summary, a risk score. It lands in a dashboard, a\nreport, an email. Then a person, if they see it at all, weighs it against\ntheir own judgement, their spreadsheet, and the way things are usually done,\nand makes the decision the way they always have.\n\nThree leaks, in sequence:\n\n- **The output is advisory.** It informs a decision at best; it does not\n  carry one. Nobody's Monday changes because a dashboard exists.\n- **The output is unowned.** No named person is accountable for acting on\n  it, so acting on it is optional, and optional actions lose to the day job.\n- **The output is unaccountable.** Nobody records what was predicted, what\n  was decided, and what happened next, so the organisation cannot tell its\n  working AI from its decorative AI. The pilot that saved nothing and the\n  pilot that saved millions produce the same slide.\n\nThe MIT research itself locates the stall at the workflow boundary, not\nthe model boundary: fluent output that did not fit how work got decided,\nquietly routed around.\n\n## What the 5% do differently\n\nThe minority that gets value does something structurally different, and it\nis visible from outside. They do not deploy AI at the organisation. They\nwire it into specific, named decisions.\n\nA decision, concretely, has four parts: a prediction of what happens if\nnothing changes, a threshold at which someone must act, a named owner who\nacts, and a record of whether the action worked. The companies extracting\nvalue put AI into that loop, not next to it.\n\nThat changes the questions. Not \"which model?\" but:\n\n- Which recurring decisions in this function are made late, blind, or not\n  at all?\n- What would a prediction have to say, and how far ahead, for the owner to\n  act on it?\n- Who is the owner, and what does their runbook say when the threshold\n  trips?\n- Where is the evidence, six months later, that acting beat not acting?\n\nNone of these are data science questions. They are operating questions, and\nthey are the reason value concentrates in so few hands: most AI programmes\nare run by teams empowered to build models, not to change who decides what,\nwhen.\n\n## The uncomfortable implication\n\nIf the gap is decisional, buying more intelligence does not close it; the\nupgraded model lands in the same dashboard its predecessor did. This is why\nthe abandonment numbers climbed in the same year model capability jumped:\nthe bottleneck moved, and budgets did not move with it.\n\nThe work that pays is less glamorous. Pick the decisions. Name the owners.\nSet the thresholds. Route the prediction to the person, not the portal.\nRecord the outcome and let the record settle which predictions deserve\ntrust. Do that for one function's ten most expensive recurring decisions\nand the P&L impact stops being unmeasurable, because measurement was built\nin from the start.\n\n*AI creates value at exactly one moment: when a person decides something\ndifferently because of it. Everything else is cost.*\n\nThis is the problem Prophesee was built for. A decision layer that turns\nenterprise data into predictions, exceptions, plans and answers, wired to\nnamed owners with the outcome recorded. Built with and proven inside global\nenterprises. If you want to see it on your own data,\n[start here](/contact).\n",1786786820103]