Between Sips #1 Marco Luciano / Everyone has AI. Now What?

An espresso with Marco Luciano from AI Studio on why buying AI tools is the easy part. Getting them to do something useful? Slightly more complicated.

Marco's coffee order: Kimbo Barista Espresso Napoli, Premium Selection, dark roast 10/13. Always an espresso. Proper Neapolitan coffee, obviously.

Reading time: about one coffee

Welcome to Between Sips, our excuse to grab a coffee with interesting people and ask them a few questions about things worth talking about. Short conversations, strong opinions and hopefully something useful to take away before the coffee gets cold.

For our first edition, we sat down with Marco Luciano from AI Studio to talk about AI strategy. Or, more specifically, why quite a lot of what companies are calling an AI strategy probably isn't one.

So Marco, what does an AI strategy look like these days?

Apparently, quite a lot like a shopping list.

"A license rollout plus a lunch-and-learn. Copilot for everyone, and then hope," Marco tells us.

It's a familiar scenario. A company invests in AI tools, gives employees access, organizes some training and waits for the productivity revolution to begin. People experiment, some become enthusiastic users and others continue doing exactly what they were doing before, only now with another subscription.

Marco isn't particularly impressed.

"That's a procurement and IT department decision. A strategy says which workflows (e.g. in marketing) changes and who owns the quality."

It's an important distinction. Giving employees access to AI isn't inherently a bad thing. It's just not the same as changing how an organization operates.

If everyone has the same tools, what actually makes the difference?

"The tech is the same for everyone, so your expertise is the only advantage left."

Marco makes a point that seems obvious once someone says it out loud. Companies spend years building expertise, developing processes and learning what works. Their best people accumulate an extraordinary amount of knowledge about customers, industries, brands and the hundreds of small decisions that separate good work from mediocre work.

Then the company buys AI software and gives it absolutely none of that knowledge – or sometimes in an unsteered way, hoping it will work out somehow.

Instead, everyone starts improvising. One person has developed an elaborate prompt library. Another types "make this sound better." Someone else has somehow convinced ChatGPT that every LinkedIn post should begin with "In today's rapidly evolving landscape."

The result? Forty different ways of using AI, forty interpretations of the brand and very little consistency.

"Most companies have decades of judgment sitting in people's heads, and none of it ends up in the AI,"

Marco explains.

And that's where the conversation gets interesting, because the answer isn't necessarily more technology.

So how do you get that expertise into AI?

Marco has a wonderfully simple way of explaining it:

"Think recipe, not improvisation."

A prompt is usually a one-off request. You ask the AI to do something, provide some context and hope the output resembles what you had in mind.

Building expertise into AI means going further. It means capturing how an experienced person actually approaches a task: what information they need, which questions they ask, what steps they follow, which standards matter and what a good result looks like.

That knowledge can then become part of a repeatable workflow, available to the wider team.

Imagine a junior marketer preparing a campaign brief. Instead of starting with a blank page and hoping they ask the right questions, they work with an AI tool that already incorporates the company's briefing process, brand guidelines and quality standards.

The AI doesn't replace the strategist. It makes some of that strategist's expertise available before the work reaches their desk.

And importantly, it doesn't mean everyone suddenly produces identical work. It means people share a stronger foundation while still bringing their own thinking, creativity and judgment to the table.

Which sounds considerably more useful than teaching 40 people to write 40 different prompts.

But is it actually making the work better?

This is where Marco gets particularly animated. Because apparently, we're still measuring AI success in some rather strange ways.

How many employees have activated their licenses? How often are they logging in? How many prompts are they submitting?

Interesting numbers, perhaps. But they don't tell you whether the work has improved.

Marco would rather measure output quality and actual changes in how people work.

He shares an example from a global industrial manufacturer where AI was integrated into campaign preparation. The team reduced drafting time by approximately 50%, saving more than three hours per person each week.

That's already a useful result. But the more interesting part was what happened to the quality of the work.

Junior team members began producing briefs that previously would have required substantial rewriting from senior colleagues. The AI wasn't simply making the same work happen faster. It was helping less experienced employees produce stronger work by making senior expertise available earlier in the process. And educating the juniors along the way.

Marco suggests a simple test: can a junior produce something close to senior-quality work on the first attempt?

That's a much more meaningful measure than the number of times someone opened Copilot this week.

Because faster mediocre work is still mediocre work. You just have more of it.

What should companies stop doing?

At this point, we're getting towards the bottom of Marco's espresso (the third sip, actually), so we ask him for one practical piece of advice.

His answer is refreshingly uncomplicated.

"Stop rolling out licenses without a single defined use case. Pick three workflows, make them excellent, then scale."

Three workflows. Not an organization-wide transformation program involving 17 committees and a very expensive PowerPoint presentation.

Start by identifying three recurring tasks where quality matters and the process can be clearly defined. For a marketing team, that might mean campaign briefing, content development or proposal preparation.

Then look at how those tasks are currently performed. Who does them particularly well? What knowledge do they rely on? What makes their output better? Where do mistakes or unnecessary revisions happen?

Capture that expertise, build it into the workflow and test whether the result is genuinely better. Not just faster, but more consistent, more useful and less dependent on one experienced person fixing everything at the end.

Once it works, improve it and expand.

Of course, none of this sounds as exciting as announcing a company-wide AI transformation initiative. But it has the considerable advantage of giving people something specific to improve.

One last question before the coffee cup gets cold. Who wins?

We ask Marco what will separate companies that genuinely benefit from AI from those that simply accumulate more tools.

His prediction is that the difference won't be access to technology. It will be how effectively companies have captured and structured their own expertise.

The organizations that get this right will have their judgment, standards and ways of working built into the tools their teams use every day. They'll be able to improve those processes, share what works and maintain consistency as they grow.

The others will have access to exactly the same AI, but without much of what makes their business different.

And that's perhaps the most interesting takeaway from our coffee with Marco.

The question isn't whether your company has AI. It's whether your AI knows anything about your company.

Otherwise, you may end up with plenty of licenses and an astonishing amount of very efficiently produced mediocrity.

At least it'll be fast.

——————————————————————————————————

Between Sips is an Agency X series featuring short conversations with people who have something interesting to say. Different guests, different perspectives and different coffee orders. Because good conversations don't always need a conference room.