Hey friends,
Someone put their hand up at the back of the room last week and asked the question I get in almost every room. "Where do we even start with AI?"
I used to give a big answer. Strategy, roadmap, maturity levels, the whole thing. I watched people nod, write it down, and then do nothing for six months. Not because they were lazy. Because the answer was too big to act on. You cannot pick up "adopt AI" on a Tuesday morning. There is nowhere to put your hands.
So now I start somewhere smaller, and it is the most useful thing I say all day. Stop trying to adopt AI. Find one boring job and kill it.
“Adopt AI” Is Still A Trap
The phrase sounds responsible. It also guarantees you never start.
"Adopt AI" has no edges. It turns into a committee, a list of tools nobody uses, a budget line, and a slow feeling that everyone else is further ahead than you. The work expands to fill the ambition, and the ambition is infinite, so nothing ever ships.
I want to be careful here, because this is where people misread me. The problem is not strategy. The problem is deliberation standing in for doing. Months of meetings about AI before a single person has used it on a single real task. That is the trap. Not the roadmap. The talking.
Meet Your ROBOTs
Here is the question I ask instead of "where can AI help?" because that question is too vague to answer.
What are the ROBOTs sucking your time?
ROBOTs stands for Repeated, Obligatory, Boring, Operational Tasks. The jobs that are already robotic in the behavioural sense. You do them again and again, you have to do them, they bore you, and they keep the lights on without moving anything forward. Writing the same status update. Reformatting the same report. Drafting the same kind of reply. Pulling the same numbers into the same shape every week.
You do not need an AI strategy to find these. You already know what they are. They are the parts of your week you would happily never touch again.
Strategy Is A Means, Not A Mountain
Now, the roadmap. We build these for organisations, so let me be straight about what it is for.
A roadmap is not the destination. It is the fastest way to find your real use cases, sequence them, and avoid wasting money on the wrong ones. Done properly it surfaces dozens of opportunities you would never have spotted task by task. In one recent strategy engagement with a manufacturer, the exercise produced more than sixty usable AI use cases across the business. You do not get to that picture by poking at one tool on a Tuesday.
But here is the rule that matters. A roadmap should take weeks, not months. The moment it becomes a year of deliberation, it has stopped being a means and started being the mountain. Same with the working group. You need one. It should be light and lean, a few people who can actually decide and act, not a standing committee that meets to discuss the agenda for the next meeting.
The order also matters more than people think. The working group should not be the first thing you do. It is something you grow into once people have seen AI work. The sequence that holds up is simple. Get the lay of the land, get hands on a real task, get a few people trained, then form the group to run the rest. Structure follows proof, not the other way around.
The Real Fork: Can You Actually Commit?
This is the part that decides everything, and it is the honest answer to "where do we start."
Ask yourself one question. Can your leadership and your team actually commit right now? Real time, real budget, real backing from the top.
If yes, do the roadmap. Lean, fast, pointed at shipping. Let it map your use cases at scale, train your people, and stand up a small working group to run it. This is the serious path and it is the better one when the organisation is ready for it.
If not yet, and this is most rooms I stand in, do not wait for permission. Do not let the absence of a grand plan stop you from starting. Pick one boring job and kill it this week. The proof you create is what earns the commitment you do not have yet. I have watched a single dead task change a leadership team's mind faster than any slide deck ever did.
Both roads end in the same place
What We’ve Actually Killed
People find this easier when they can see the shape of it, so here is a list to steal from. Some of these are ours, some are the kinds of jobs we take off teams every week.
Bid and tender writing. We took our own from six or seven days down to about one. Same quality, a fraction of the time.
Quote generation. A manufacturer drowning in manual quotes now turns them around in a fraction of the time it used to take a person.
Site and incident reporting. A voice-first agent that turns a spoken update on the floor into a clean, compliant written report, so the human never opens a form.
Weekly status updates. The same report, in the same shape, written for you from the same source every week.
First-draft replies. The repetitive emails and messages that follow a pattern, drafted for a human to approve in seconds instead of writing from scratch.
Pulling numbers into the same shape. The recurring spreadsheet wrangle that eats a morning and changes nothing.
Code boilerplate and documentation. For the dev teams, the repetitive scaffolding and the documentation nobody wants to write.
None of these needed a transformation programme. Each one is a single ROBOT, taken off a human, this week.
The Dumb Monkey Move
The mistake is confusing motion with progress. Buying the licences, forming the big committee, commissioning the year-long strategy, talking about AI in every meeting. It feels like adoption. It is just expensive standing still.
The smart move is smaller and slightly embarrassing in how simple it is. If you are ready, run a lean roadmap and start shipping. If you are not, take one boring job off your plate by Friday. Either way, something real happens this week. That is the whole move.
Your Ten Minute Version
Try this before the week ends.
Write down every repeated boring task you did this week. Do not filter, just list them. Now circle the one you did the most times. Frequency beats glamour, so resist picking the impressive one. Give only that single task to AI this week and get it to a standard you would actually sign off on. Leave everything else on the list alone.
You are not trying to do all of it. You are trying to do one of it. Completion beats coverage every time.
A Calm Takeaway
You do not transform an organisation by adopting everything at once. You also do not transform it by refusing to plan. You transform it by getting your hands on one real use case, proving the time saved is real, and then doing it again with a little more structure each time, until one day you look up and the shape of the work has changed underneath you.
Strategy has its place. So does a working group. They just have to serve the doing, not replace it. The pressure to "do AI properly" before you do any of it is the thing keeping most people stuck. Let that go. Pick the boring job you do most, or run the lean roadmap if you are ready. Then start. See you on the other side of it.
See you next week,
— Aamir
📲 Resources & Links
🎧 Listen to the Podcast Episode on: Spotify | Apple Podcasts | YouTube
📘 Book: The CEO Who Mocked AI (Until It Made Him Millions) by Aamir Qutub