Work Redesign Is a Practice, Not a Project
Companies usually redesign work around AI as a project, with a start date, a finish date and a deliverable. But the business keeps changing, and so do the tools, so a one-time redesign soon goes out of date.
The Eagle Hill Consulting survey I cited in Why Most AI Rollouts Stall found that only 45 percent of leaders at large U.S. companies already using AI say their company has an established practice for continuously reviewing and improving work.¹ AI adoption and continuous improvement are not the same thing, but they belong together. Every review of how work gets done should now ask where AI fits, and every AI tool needs the same regular review as the process around it.
Why does a one-time redesign become out of date? Partly because the business changes: people leave, customers change and workarounds creep back in. Partly because the tools change. MIT Sloan researchers found that one step AI handles poorly can break an otherwise automated chain of tasks.² AI tools change every few months, and the change cuts both ways. A step that broke the chain last spring may work today, and a step that worked may break when a new model is released. Only a review will tell you. Meanwhile, new uses your people discover often go no further than the work on their own desks, as I described in Someone in Your Company Has Already Figured This Out.
A standing practice does not need to be elaborate. It needs five things:
An owner, typically an operations manager or controller.
A review of one process at a time, monthly or quarterly.
A permanent place on the management meeting agenda.
The same question at every review: which steps exist only because of a limit we no longer have?
A simple way for people on the front line to bring an idea to the person able to act on it.
Simple does not mean improvised. Business process mapping, Lean (a method for stripping out wasted steps and waiting) and similar methods, refined over decades in manufacturing, give each review a proven sequence: map how the work actually happens, find where it waits or changes hands, fix the cause, and measure the result. Someone trained in them, whether on staff or brought in, will find more in one review than a group starting from a blank page.
This takes management time, and the return builds gradually as people learn what to look for. Bain & Company sees the same pattern in companies pulling ahead with AI, which design their approach so “each deployment makes the next one smarter, faster, and cheaper.”³
Look back at your management meeting agendas for the last three months. Count how many times someone reviewed how a piece of work actually gets done. If the answer is zero, put one process on next month’s agenda. The practice starts with that first review.
Sources
¹ Eagle Hill Consulting, “New Eagle Hill Consulting research finds AI is reshaping how organizations work, but leadership and culture lag behind,” September 8, 2026. Survey conducted by Ipsos, August 3-6, 2026.
² Kristin Burnham, “How AI is reshaping workflows and redefining jobs,” MIT Sloan Ideas Made to Matter, April 22, 2026, reporting on “Chaining Tasks, Redefining Work: A Theory of AI Automation” by Peyman Shahidi, Mert Demirer, John Horton, Nicole Immorlica and Brendan Lucier.
³ Sarah Elk, Chuck Whitten, Hernan Saenz, Gene Rapoport, Nicolas Bloch, Anne Hoecker and Pascal Gautheron, “Proprietary Intelligence: How to Win with AI,” Bain & Company, June 25, 2026.