More and more of the leaders we work with have moved past the first stage of AI. Their teams use it every day to draft, summarise and analyse. The question they are bringing to us now is harder: what should our workflows actually look like now that AI is part of the work? If we were using AI from the start, would the way our business runs be different?

Most organisations have started in a sensible place. They took the processes they already had and added AI to the steps inside them. The report gets drafted faster, the ticket gets summarised, the code review comes with suggestions. That’s a fine way to begin, and while there are visible benefits, they are usually modest, because fundamentally everything around those steps stays the same. You can’t sprinkle AI on top of a process and expect it to change how the business runs.

A lot of workflows were designed to manage friction, mostly the effort of getting knowledge from the person who has it to the person who can turn it into something usable. When that friction goes, some steps are left with nothing to do.

What happened when an analyst left

I saw this first-hand while working alongside a client team. Their full-time analyst left, and the conversation turned to how to replace them. While that conversation was happening, a domain expert in the business had started producing the customer reports directly with AI. Work that had taken the analyst weeks each cycle now took the expert minutes.  The reports were often sharper, too, because the person writing them knew the customers and the audience. 

Pulling the data, shaping the narrative and making it presentable had all sat between the person who knew what was important to their customer and the finished report. Once the expert could do all of that themselves, most of the handoffs and the waiting in that workflow had no reason to exist.

What these workflows can look like

The same pattern shows up well beyond reporting: between an idea and a prototype, a business problem and working code, or a customer conversation and a tested proposition.

When the person who understands the problem can get close to a working output themselves, the workflow gets shorter. The design effort moves somewhere else: where the data comes from, where a person reviews the output, what happens when something goes wrong, and how you’d know if it had.  That’s different work from tidying the boxes on a process map, and most organisations haven’t done much of it yet. The ones getting the most out of AI have. In ++McKinsey’s 2026 State of AI survey++, “nearly three-quarters of high performers report fundamentally redesigning workflows because of their AI use”, up from 55 percent the year before.

What they should look like is the harder question

When I look at a workflow now, I try to separate the steps that exist because of friction that has gone from the steps that carry judgement. The first kind can often be removed. The second kind usually deserves more time than it ever gets.

In the reporting example, the judgement was knowing what mattered to the customer. That now sits with the person who has it, and the steps that only moved work between two people have gone. That is what I think these workflows should look like: built around the people who hold the judgement, keeping them as close to the output as possible, with a review point wherever a mistake would really matter. 

I opened my talk at the ++DIGIT AI Business Summit++ in May with a line I still believe: “You are not managing a transformation; you are managing an inflection point.”  A transformation makes the existing process better. At an inflection point, some of the process stops making sense, and the challenge becomes deciding what replaces it.

Where to start

Pick one workflow your team runs every week and write down every step. Next to each one, note whether it does the work or moves the work between people. The steps that only move work around are often designed for friction that isn’t there anymore, and that is where I would start.

If you are exploring similar questions, we are always happy to compare notes.