THE VALUE GAP · No 06
The switch-off test
There is a one-question test that separates an AI tool from an AI transformation. Turn it off tomorrow, and see what stops.
———
Here is a test you can run on your own organisation without a consultant, a survey or a steering committee. Imagine that tomorrow morning, every AI system you have deployed is switched off. The copilots go dark. The agents stop. The models return nothing. Now ask one question: what actually stops working? Not slows down — stops. What could your organisation no longer do at all?
For most enterprises, honestly answered, the reply is uncomfortable: nothing stops. Things get slower, a little more manual, a little more expensive. People go back to doing tasks they had recently handed off. But the business runs. The workflows are the same workflows; the org chart is the same org chart; the way decisions get made is unchanged. The AI was making the existing machine turn faster. It was not the machine.
The World Economic Forum, with Kearney, has just put a name to what that test reveals. Drawing on more than fifty of the world's most advanced AI enterprises, they sort organisations into three kinds. AI-enabled: AI is applied to discrete tasks on top of existing workflows — remove it and nothing structural changes. AI-first: workflows, roles and decision rights have been redesigned around AI — remove it and the business cannot run as designed. AI-native: AI is the product. Their litmus test is exactly the one above: if the AI were removed, could the business still operate? If the answer is yes, you are AI-enabled — and being AI-enabled is a ceiling on what you can ever get back.
If turning it off changes nothing structural, you bought a tool, not a transformation.
We have watched this film before, and the Forum is candid enough to point at the reel. When factories first electrified, the early adopters pulled out the steam engine, dropped an electric motor into the same spot, and left the building exactly as it was. Their energy bills fell — by twenty to sixty per cent — and their output did not move at all. The step-change came years later, when pioneers like Ford stopped treating electricity as a replacement part and redesigned the entire plant around what it now made possible: power distributed to every workstation, the line resequenced, the whole blueprint rebuilt. That redesign, not the motor, created the modern factory.
AI is the same class of technology, and most organisations are still at the motor-in-the-old-factory stage. The numbers say so plainly. Despite more than 250 billion dollars poured into AI globally in 2025, only a quarter of companies say it is having a transformative effect — and eighty-four per cent have not redesigned a single job around what AI can now do. They have electrified the workflow and left the building standing exactly where it was.
This is the value gap, viewed through a light switch. The gap has never been about whether the technology works; it plainly does. It is about whether the organisation was redesigned to capture what the technology makes possible, or merely had the technology bolted onto processes it should have retired. The switch-off test cuts straight to that question. An organisation that would grind to a halt without its AI has rebuilt itself around intelligence; the value is structural, and it compounds. An organisation that would barely notice has bought capability and called it transformation — and it will keep paying for that capability, quarter after quarter, while wondering why the return never arrives.
They electrified the workflow and left the building standing exactly where it was.
None of this is an argument against being AI-enabled as a starting point. Every organisation begins by bolting the motor on; that is how you learn where the power is useful. The mistake is stopping there and mistaking the efficiency for the transformation — treating a faster version of the old process as though it were a new business. The switch-off test is not a gotcha. It is a compass. Run it honestly, and it tells you the one thing the vendor demos never will: whether you are optimising the thing you already had, or building the thing you could become.
If the answer is that nothing much would stop, that is not a failure. It is a map. It tells you exactly where the redesign has not happened yet — and where the value is still waiting to be built.
———
THE VALUE GAP — The full argument, and the BRIDGE framework for redesigning the work rather than bolting AI onto it, is in Bridge The Value Gap, out now at rodneyhobbs.com.
———
References
1. World Economic Forum with Kearney, The AI-First Operating System: A Blueprint for Operating and Business Model Innovation, June 2026 — the AI-enabled / AI-first / AI-native archetypes and the "if AI were removed, could the business still operate?" litmus test.
2. WEF/Kearney, 2026, citing the electrification of manufacturing — electric motors substituted into steam-era layouts delivered ~20–60% energy savings but no productivity gain until factory blueprints were redesigned (e.g. Ford, 1919–1926).
3. WEF/Kearney, 2026 — over US$250B invested in AI globally in 2025; only ~25% report a transformative effect; 84% of companies have not redesigned jobs around AI capabilities (Deloitte State of AI in the Enterprise).