Hey friends,

Lately I keep reading the same job ad. AI Expert. AI Consultant. AI Developer. A business has decided to get serious about AI, so they've posted a role for it. Then I read the description, and it's clear they have no idea what they're actually looking for. The title sounds confident. The body underneath it is a wish, not a job.

Here's what happens next. The post pulls in a stack of applicants with traditional machine learning backgrounds, because that's who the word "AI" still attracts. And if the business can't afford a senior person, they hire a fresh graduate and hope the capability magically appears. It almost never does. Not because those people aren't capable, but because nobody decided what the role was before hiring for it.

That's the real problem. Not whether to hire. The order you hire in, and the fact that most people start by hiring at all before working out who they need first.

Building this capability, whether you're a fifteen-person business or a fifteen-hundred-person one, comes down to filling three roles. And those roles have an order. Fill them in the right sequence and the whole thing builds on itself. Post a vague ad and hire whoever it attracts, and even a strong hire sits there doing very little.

Most leaders don't have a hiring problem. They have a sequencing problem, dressed up as a job posting.

When a business posts "AI Expert" without defining it, the applications sort into two types, and both are the wrong place to start.

The first is someone with a traditional machine learning background, and on a tighter budget, that usually means a fresh graduate. Most university AI programs are built around traditional AI: machine learning, predictive models, building and training algorithms from the ground up. Real, valuable work. Almost never the work your business needs right now. What you actually need is someone who can take models that already exist and use them to build agents and orchestrate workflows around your business. Different skill, and you won't find it in a course syllabus.

The second is a technical person hired to "build the AI." This feels safe. Surely a hard technical problem needs a hard technical hire. But building an AI agent or workflow is only about twenty percent technical. The other eighty percent is understanding the business, the process, and designing a system around how the work actually gets done. Hire purely for the technical twenty percent and you've bought a beautifully built answer to a question nobody framed properly.

Both make the same mistake. They put the building first. And building is not where this starts.

A business that runs well on AI needs three roles filled, not three new hires.

First, the AI Champion, driving adoption across the organisation, helping people, turning curiosity into real use. Second, the AI Builder, actually building the systems. Third, the AI Process Owner, or agent owner, who owns the work the AI is doing.

Most people start at the second one. You start at the first.

Without a champion, your builder is useless.

The champion drives adoption. They help people, run the experiments, translate between the business and the technology. If you as the leader can play this role yourself to kickstart things, do that. If your team is too buried to spare anyone, bring this person in, and bring them in with some authority.

But be clear on what the role is and isn't. This is not a technical role. It's a people person with genuine business acumen and a feel for where the business is heading, who also happens to be a nerd by choice. That combination is rare, and getting it right is the single most important move in the whole effort. Depending on your size, you may want several champions, not one.

Throw out the normal interview. You almost want a live demo instead.

First, ask the candidate to explain AI, and agentic AI, in simple terms. Then ask them to explain the benefits and risks of AI using no technical words at all. If "LLM" or "model" comes out of their mouth, they've failed. The whole point of a champion is that they can carry this to people who will never speak that language.

Then flip it. Have someone on the panel describe their own role and day-to-day work, and ask the candidate to pull the AI use cases out of it in real time. Then ask them to repeat back what they understood about that person's job and how it gets done. In a single meeting you'll see exactly how well they can listen to a business problem, understand a workflow, and find where AI fits.

One warning. I'm tempted to say "just hire a business analyst," but be careful. A lot of business analysts came up in the software world and are wired to extract software requirements, not business requirements. You want someone hunting business outcomes, not building a spec sheet.

Once you have a champion, the builder makes sense. This is the role most people think they're hiring for, and the one role you genuinely do hire for. With a catch.

You won't find the full skill set ready-made, not from scratch, not for the next couple of years. So you have a few honest options. Upskill an existing technical person who also has business acumen. Bring the role in from an AI agency as a contractor who leads the build. Find a freelancer on Upwork who's been selling AI automation services and bring them in-house. Or outsource the building entirely to an AI agency.

Most of those are renting, not owning. That's the right instinct for now. You rent the builder once the champion is in place to point them at the right problems. Not before.

The third role is the one that quietly makes everything stick, and you never advertise it.

Your agent owners already work for you. They're the people who own the process today: the accounts person who owns invoicing, the ops lead who owns onboarding. They don't get replaced by the agent. They get upskilled to own the agent doing their work. A redeployment, not a recruitment. And the job of training them sits with your champions, which is why the owners can only be grown once a champion exists to grow them.

Not every business can fill all three at once, and you don't need to. The sequence just compresses to fit what you've got.

If you're a small business, your champion can be you. You kickstart it, and you hire for the builder. If you can find a real champion and a real builder, bring both in.

If you can only afford one hire, hire the champion first and outsource the build. The champion holds the whole thing together; the build can be rented.

And if you're not ready to invest at all, bring in an AI agency, but be specific. Don't hire one just to hand you a roadmap. Hire one that builds the roadmap and then builds the thing with you. Look for an agency that can execute a roadmap in one to three weeks, gives you six to eight hours of training so you can grow your agent owners inside, and helps you nurture a champion inside while they handle the build. That leaves you stronger, not more dependent.

One warning before you start, because I've watched it cost people. Do not get lured by off-the-shelf AI software.

You already know this pain. You've spent years wrestling with systems that don't talk to each other. The last thing you need is a pile of third-party AI agents that each do one trick and none of which work together. Same mess, new coat of paint.

What you're building is not a drawer full of AI tools. It's an AI operating system, one coherent thing where the agents, the people, and the processes connect. The three roles, in order, are how you build that system. Tools are what you buy when you don't have one.

Building AI capability isn't a hire. It's an order of operations. Champion first, because nothing works without someone who can translate and carry people along. Builder second, rented until the skill is easier to find. Owners third, grown from the people who already do the work.

So before you write a job ad, don't ask "who's the most impressive AI person I can hire." Ask "who comes first." The answer is almost never the one you were about to reach for.

One genuine offer. If you want help shaping the job description or interview questions for any of these roles, reach out. And if you're hunting for the right person, I might be able to point you to someone, though the pool is small.

See you next week,

— Aamir

🎧 Listen to the Podcast Episode on: Spotify | Apple Podcasts | YouTube

📘 Book: The CEO Who Mocked AI (Until It Made Him Millions) by Aamir Qutub

🚀 Dumb Monkey AI Academy

📱 Dumb Monkey AI Academy App: Apple | Android