The AI Job You Actually Have

The majority of AI leaders I’ve worked with over the past five years never applied for their job.

At some point, they just found themselves in charge of "the AI topic". Typically without extra headcount, or budget. But with leadership expectations of showing some "updates" next quarter.

I've seen this "+1" responsibility across every industry I work in — always without anyone initially explaining what success looks like. And as you can imagine, "doing something with AI" isn't a strategy. (Very similar to how "Get some valuable insights" isn't a good directive for a Head of Data).

What I've learned is that it typically boils down to one of three core jobs. And figuring out which job you’re really doing helps tremendously in getting the first steps right.

Let’s dive in!

The Three Jobs

Most AI mandates I’ve encountered could be put into one of three buckets. However, leadership rarely tells you which one you're in – often they don't even know themselves.

But each requires a fundamentally different adoption approach.

Bucket 1: The ROI Job

This is the most straightforward job – and the most unforgiving. Leadership wants more money. You need a specific metric, a realistic target, and a fast path to proving value

What leadership actually wants: Quick margin impact. Measurable cost savings. Short-term revenue gains. Numbers that show up on the next P&L.

What you're evaluated on: Did this make or save money? What’s the best investment of our resources?

I worked with a services business in a commoditized market. Their industry was in a race to the bottom, and leadership needed AI to cut operational costs — not eventually, but within two quarters. There was no appetite for experimentation or "learning". Either the number moved or the project was a failure.

What I’ve seen work here is to find one workflow where AI creates measurable financial impact or at least moves a clear KPI and and ship it fast. Start narrow. Accept 80% accuracy with human backup. Prove the math before you scale.

The trap to avoid here is to build impressive demos that never hit production. Discovery feels like progress, but delivery is what gets measured.

Bucket 2: The Table Stakes Job

This job is less about ROI and more about risk management. Leadership isn't expecting you to reinvent the wheel, they're expecting you to not embarrass the company.

What leadership actually wants: Competent (!) execution of the "obvious" things. Not looking stupid compared to competitors, partners and shareholders.

What you're evaluated on: Are the basics covered? Are we keeping up with the market?

I've seen this with a financial services firm in a slow-moving industry. No one was really doing anything revolutionary with AI, but everyone was watching everyone. The mandate wasn't "transform the business" — it was "make sure we have a credible answer when the board asks what we're doing about AI."

So that’s a Table Stakes role. What works here is to identify the 3-4 "obvious" AI applications for your industry and have a clear plan for each. No one’s asking you to be extra "smart" or "innovative" here. You don't need a moonshot. Just a sensible roadmap that shows you're paying attention and you’re able to execute within a reasonable timeframe and budget.

Pitfalls include over-engineering solutions for problems that just need a competent implementation. Trying to be Category Leader when leadership only wants Table Stakes will just burn credibility and budget.

I’ve seen this play out before during the peak data hype cycle where every CEO demanded to be "data-driven", but what they actually wanted was working dashboards.

Your strongest enemy is your ego.

Bucket 3: The Category Leader Job

This is the most fun, but also the most challenging job because you can't just point to cost savings or checkbox implementations. You need a strong vision – and a credible path to get there.

What leadership actually wants: Bold positioning. Competitive differentiation. A vision that makes investors, customers, or the board believe you're ahead of the curve. (Ideally by far!)

What you're evaluated on: Is this ambitious enough? Does it position us as leaders? Is it convincing?

What matters here is that the completeness and cohesiveness of the vision is actually more important than the actual probability of realizing it.

Think Elon wanting to build a multi-planetary species and starting with building reusable rockets as a step 1. Will we eventually get there? No idea, but reusable rockets are still useful.

I've worked with a company whose majority shareholder was a tech firm. The expectation here wasn't incremental improvement, but something along the lines of "let’s push the boundaries of what's possible in this industry". Of course, ROI still played a big role (someone had to fund this whole mission), but pure ROI math wasn't enough. "Keeping up" wasn't enough. The job was to take the lead.

So if you're in a Category Leader job, this one requires defining what AI-native looks like for your industry and build a roadmap to get there. Draw a picture of where your organization will be in 2-5 years from now. And what needs to be done right now. You'll need executive buy-in, a tolerance for longer payback periods, deeper pockets and the ability to tell (and update) a compelling story while you're still building.

This role requires more storytelling than technical skills – and I don't mean that in a bad way. You need to have good tech enablement to make this happen, but the payback will be so distant that you need someone to buy into the story first. At some point, you have to prove you can execute — not just strategize.

How to Figure Out Which Job You Have

Leadership won't always tell you directly. Sometimes they'll say "ROI" but actually mean "just don't embarrass us". Sometimes they'll say "innovation" but actually mean "cut costs". You have to read between the lines.

Two questions I always ask early:

"What's your North Star?" — This separates ROI jobs (specific metric) from Category Leaders (market position) from Table Stakes (staying competitive).

"What gets leadership genuinely excited?" — The answer is often more revealing than the official mandate. One client told me cutting headcount was off the table – but not having to rehire when people left? That got enthusiasm. Because it was what everyone else was doing. And suddenly we were in a Table Stakes Job.

Once you know which play you're in, everything else gets simpler: what to prioritize, how to measure success, what to say in executive updates. Profitable AI looks different depending on which job you're doing.

The key is making sure you and leadership are measuring success by the same standards.

A (Free) Course for This

I built a LinkedIn Learning course for exactly this situation – turning a vague AI mandate into business-aligned execution. It covers prioritizing use cases, proving value early, getting stakeholders on board, and scaling without derailing your data strategy.

You can watch the course for free using this link:

Conclusion

What's helped me most in any of these situations: getting alignment before getting busy.

Conversations with leadership are worth having early and often. The goal is to turn mutual assumptions into clear expectations. The two questions above will help you get started – and hopefully save you months of wasted effort.

Keep going!

See you next Saturday,
Tobias