Change management assumes the work is right; AI sometimes changes what the work should be. The Reinvention Threshold uses two questions and four responses to sort each task in a fortnight.
Most AI adoption programmes are run as change management, and most of them are aimed at the wrong problem. Change management assumes that the work is right and the people need to be brought along to a new way of doing it. That assumption holds for a new finance system. It holds far less well when the technology alters what is cheap, what is slow and what is possible, because then the work itself may need to be redesigned.
The distinction matters commercially as well as intellectually. An organisation that applies a change playbook to a reinvention problem spends its budget on communication, training and champions, and ends with staff who use AI faithfully to reproduce a process that no longer makes sense. It looks like adoption. It delivers little value.
Change management asks how to get people to do the work differently. Reinvention asks whether this is still the work.
We use a short diagnostic, the Reinvention Threshold, applied task by task rather than organisation-wide. It rests on two questions.
The first is about purpose: is the outcome this process was designed to produce still needed, in its current form? A monthly narrative report may exist because a funder wanted assurance that activities were on track. If the funder would now accept a shorter structured update, the purpose has shifted.
The second is about constraint: is the limitation that shaped the process still binding? Many workflows carry the fingerprints of an old constraint. Reports are long because drafting was slow. Translations are batched because they were expensive. Reviews are sequential because parallel review was hard to coordinate. When AI removes the constraint, the shape it created may no longer serve anyone.
Answering the two questions places each task in one of four responses.
Most organisations find that a substantial share of what they call AI adoption is Adapt or Rebuild work being managed as Adopt.
Change management playbooks are comfortable for understandable reasons. They are familiar, they produce visible activity, and they place the burden on staff behaviour rather than on the design of the work. Leaders can run them without asking anyone to question a long-standing process, which is politically easier.
Reinvention is harder because it requires someone with authority to say that a process people have owned for years might not be needed. It exposes trade-offs between roles, and it takes longer to show results. That is why it tends to be deferred, and why organisations that defer it often find that a competitor, a peer or a funder has redesigned the expectation for them.
Adoption without redesign is the most expensive way to make a legacy process slightly faster.
Reinvention can be used as an excuse for disruption. It is not a mandate to redesign everything, and it is not a claim that people's jobs should be reduced. The threshold is deliberately demanding: a task should move out of Adopt only when there is a clear answer to why. Tasks involving safeguarding, beneficiary trust or legal accountability warrant particular care, because a process that looks slow may exist precisely to protect something.
Consider an illustrative example. A small foundation sends a narrative report to its board every month, twenty pages long, assembled over four days by three people. Asked what the report is for, the board chair says the board reads the two-page summary and the risk table. The purpose is oversight, and the constraint was the effort of drafting. Once AI can assemble the detail on request, the sensible design is a one-page dashboard monthly, with the full narrative produced on demand.
Nothing about that outcome required a training programme. It required someone to ask the purpose question, and someone with authority to act on the answer. That is the reinvention conversation in its simplest form, and it is the one most organisations never have.
Use change management where the work is right and reinvention where the work itself has been overtaken. The Reinvention Threshold gives a fortnight's diagnostic for telling them apart, and it protects you from paying for adoption that only accelerates the wrong process.
Ask each department to list its ten most time-consuming recurring tasks. For each, one person writes two sentences: what the output is for, and why the process has its current shape. Then a small group, including someone from outside the department, sorts each task into the four responses.
Two things usually happen. Most tasks land in Adopt, which confirms the standard programme is right for them. A handful land in Adapt or Rebuild, and those are where the value lies: give them proper design attention, with the affected staff in the room, before any tool is rolled out. The exercise also tends to reveal disagreement about purpose. When two people in the same team give different answers to "what is this report for?", you have found something more useful than any efficiency gain.
No external statistics cited. The Reinvention Threshold is a framework proposed by M.A.I. Consulting; the foundation example is illustrative.
Why AI adoption decisions should be made task by task rather than for a whole job or department, and how a task inventory replaces the blanket yes or no.
Team AI Training · 4 minGuide · 20 September 2026How to break a task into steps and write down the handover point, usually just before the first judgement call, where AI output ends and human responsibility begins.
Team AI Training · 4 minAnalysis · 5 October 2026Why do AI pilots impress and production disappoint? Because most pilots are built to show it can work, not whether it will. The Three-Gap Test: material, people and ownership.
Custom Agents & Tools · 6 minIf this is the question on your desk, a thirty-minute call tells you whether the service fits, or that you do not need us yet.