Every leader I speak to right now is under pressure to “use AI”. Clients ask about it, competitors talk about it, and teams are experimenting on their own. Some of that pressure is healthy. It forces us to question old habits and look hard at how work actually moves.
But it also pushes businesses towards tools before they have understood the work those tools are meant to improve.
The better starting question is not “which AI tool should we buy?” It is “where is our work losing time, or quality?” That shift sounds small. It changes everything, because it moves the focus away from technology for its own sake and towards the places where AI genuinely helps people do better work.
The risk nobody talks about
Most businesses are worried about falling behind. That fear is real, and doing nothing is a legitimate risk.
There is a second risk that gets far less attention: rushing into AI without redesigning the work around it. If a process is unclear, AI will not make it clear. If the data is messy, AI will not make the decision better. If approvals are already slow, bolting a tool on top makes the handover more confusing, not less.
So, the most useful AI opportunities are usually unglamorous. They sit in reporting that takes too long, client information scattered across five systems, repeated admin that pulls people off higher-value work, and briefs that lose context every time they change hands. These rarely make headlines. They are exactly where teams feel the daily drag.
Take a simple one. A weekly status report that used to swallow the first hour of someone’s Monday, pulling numbers from three places by hand. The fix was not “add AI”. It was redesigning the flow, so the data lives in one place, letting AI assemble the first draft, and keeping a person to sense-check it before it goes out. The hour became ten minutes. The judgement stayed human. That is the pattern worth repeating, and it starts with understanding the task, not automating it on instinct.
Design the human layer, not just the tech
Most AI conversation fixates on the technology: models, prompts, agents, integrations. That matters, but it is only half the picture.
The other half is the human layer, and it needs designing just as deliberately. Who reviews the output? When should AI be used, and when should it not? What level of accuracy is acceptable? Where does accountability sit when something goes wrong?
These are not blockers. They are what separates faster output from better work. A tool can generate a result. It cannot decide whether that result is good enough, right for the client, or safe to act on. That judgement still belongs to people.
A word of balance, though. The opposite failure is just as real: endless discovery, mapping every process to perfection, and never shipping anything. You do not need a finished strategy to begin. You learn most by putting one redesigned workflow into real use and watching what it teaches you.
So if you are looking at AI now, start by mapping how work actually happens, not the tidy version on paper. Find the decision points where better context would help. Spot the repetitive work that still needs quality control. Get your data into a state people can trust. And be clear about where humans enter the loop and what they own.
“Work smarter” gets said a lot. In practice it should mean something specific: less time hunting for context, less repeated admin, cleaner reporting, clearer handovers, and more room for the judgement, creativity and relationships that actually move a business.
That is the part I find most interesting about AI. Not the novelty of the tools, but the chance to redesign how work moves.
The businesses that get this right will not just use AI to go faster. They will use it to build better habits.
AI should never be a shortcut around people. It works best as a support layer around better work. AI-fed. Human-led.
