You've heard the sentence in some form for two years now. AI is going to change your job. It's in your CEO's all-hands. It's in the webinar your VP forwarded. It's in the LinkedIn post a peer wrote that got 400 likes.
Here's what nobody followed it up with: how.
If you're in sales or marketing, you're being squeezed from two directions at once. From above, leadership now expects AI productivity gains — they've read the same reports, and they want the numbers. From the side, you can feel that some of your peers and competitors are already moving faster: researching accounts in minutes, turning around content in hours, running follow-up that doesn't slip. And in the middle is you, with a ChatGPT tab open, using it to clean up an email — wondering if that's it, or if you're missing something.
You're missing something. But probably not what you think.
The divide in 2026 is not between professionals who use AI and professionals who don't. That war is over — almost everyone uses it. The real divide is between the people who use AI for one or two shallow tasks and the people who have quietly rebuilt how they work around it. This piece is about that line, why so few people have crossed it, and how you can — from any source you like.
01 The adoption gap
Almost everyone "uses AI." Almost nobody uses it well.
Start with the uncomfortable number. In EY's 2025 Work Reimagined study — 15,000 employees and 1,500 employers — 88% of people said they use AI at work in some form. But only about 5% qualified as advanced users who combine multiple tools into real workflows. Everyone else clustered around the basics: roughly half use it to look things up, about a third to summarise a document. (EY, 2025)
Adoption is nearly universal. Real use is rare.
This is the trap, and it's a comfortable one: drafting emails, fixing grammar, rewording a paragraph, translating a message. These are genuinely useful. I'm not going to pretend they aren't. But notice what they have in common — they polish work you were already going to do. They don't change your output. They make the same email slightly faster and slightly smoother. That's the shallow end of the pool, and most professionals are standing in it, water at their ankles, calling it swimming.
It can even cost you. Microsoft's research team reported that around 40% of employees had received "workslop" in the past month — AI-generated content that looks polished but is inaccurate or useless, so someone downstream has to redo it. (Microsoft Research, 2026) Shallow use without judgement doesn't just fail to help. It moves the work, it doesn't remove it.
of employees say they use AI at work in some form.
EY 2025
qualify as advanced users running real AI workflows.
EY 2025
received AI 'workslop' — polished but useless output — in the past month.
Microsoft Research 2026
So what does "using AI well" actually mean? Not a better autocomplete. It means treating AI less like a search engine and more like a junior team member you can hand an entire piece of work to — research this account end to end, draft and personalise this sequence for these forty prospects, pull this campaign's numbers apart and tell me what moved. That's a different relationship with the tool. Nobody arrives at it by accident.
02 Why the gap exists
The gap isn't the tools. It's that nobody taught you.
Here's the part that should take the pressure off you a little.
In the Microsoft and LinkedIn Work Trend Index — 31,000 people across 31 countries — 75% of knowledge workers were already using AI at work. But only 39% of them had received any AI training from their company, and only 25% of companies even planned to offer it. So people did the obvious thing: 78% brought their own AI tools to work with no guidance at all — and at small and mid-sized businesses that figure was 80%. (Microsoft & LinkedIn, 2024)
Sit with what that means for you specifically. You were handed a tool and a vague expectation. No structure. No path. The most powerful general-purpose technology to land on your desk in a decade, and the training plan was "figure it out." Most professionals are self-taught on AI because there was no other option — and self-taught, with no feedback loop, is exactly how you stay at the shallow end without noticing.
This is also why the usual advice doesn't work. "Just google a good prompt." "Watch this YouTube playlist." You can learn a clever prompt that way, the same way you can learn one chord on a guitar. What you can't build that way is a system — a connected set of moves you run every week, refine, and trust. Prompts you can google. Fluency you have to practise, ideally where someone can tell you what you're doing wrong.
None of this is a you-problem. It's structural. But structural problems still have your name on the outcome.
03 What fluency looks like
What "AI-fluent" actually looks like
Let's make the other side of the line concrete, because "fluent" is a word that means nothing until you see it.
In the same Microsoft and LinkedIn research, the small group of genuine power users hadn't just adopted a tool — they'd reorganised their day around it, saving 30-plus minutes daily, with more than 90% saying it made an overwhelming workload manageable. (Microsoft & LinkedIn, 2024) The difference between them and the 88% wasn't a secret tool. It was practice and structure.
What does that look like in sales? Bain's 2025 analysis found reps spend only about a quarter of their time actually selling — the rest goes to research, admin, and reporting — and that AI, used properly, can roughly double that active selling time. HubSpot's surveys put the same point bluntly: most reps get only around two hours a day in front of buyers. (HubSpot, 2025) An AI-fluent rep doesn't use AI to "write a better cold email." They compress account research from 45 minutes to five, generate genuinely personalised outreach at volume, prep discovery calls in minutes, capture notes automatically, and let follow-up run on rails. ZoomInfo's 2025 survey of more than 1,000 go-to-market professionals found that those using AI weekly reported being about 47% more productive and getting back roughly 12 hours a week. (ZoomInfo, 2025)
productivity for sales pros using AI weekly.
ZoomInfo 2025
given back per week to sales AI power users.
ZoomInfo 2025
productivity for marketing pros using AI weekly.
ZoomInfo 2025
What does it look like in marketing? In that same ZoomInfo study, marketing AI users reported around 44% higher productivity and roughly 11 hours back per week — not from one tool, but from running research, content production that holds brand voice, campaign analysis, and lifecycle automation as connected workflows rather than one-off prompts.
See the pattern. In both cases, the transformation is not cosmetic — a smoother sentence here, a faster draft there. It's structural. It's a chain: research feeds drafting, drafting feeds personalisation, personalisation feeds outreach, outreach feeds analysis, analysis feeds the next round. The shallow user improves one link. The fluent user owns the whole chain.
And this is not a sales-and-marketing curiosity. McKinsey's 2025 State of AI survey of nearly 2,000 organisations found AI in use across most business functions, with the deepest value concentrating in exactly these two — marketing and sales. (McKinsey, 2025) This shift is happening across every industry at once. It is simply arriving at your desk through the work you already do.
04 Why this matters to you
Why this is about to matter for you personally
It would be easy to read all of this as someone else's problem — the company's, the future's. It isn't, and the data is fairly direct about why.
The World Economic Forum's Future of Jobs Report 2025, built on more than 1,000 employers, projects that 39% of workers' core skills will change by 2030, names AI as the single fastest-growing skill in the entire survey, and reports that 77% of employers plan to reskill their workforce for AI — while 63% already call the skills gap their biggest barrier to transformation. (WEF, 2025) On the hiring side, the Microsoft and LinkedIn research found that two-thirds of leaders wouldn't hire someone without AI skills, and 71% would rather take a less-experienced candidate who has them over a more-experienced one who doesn't. (Microsoft & LinkedIn, 2024)
And the floor keeps rising. Gallup's tracking shows frequent AI use among white-collar workers in the US nearly doubled in two years. (Gallup, 2026) "I use ChatGPT sometimes" is quietly becoming the new "I know a bit of Excel" — true of nearly everyone, and therefore worth almost nothing on its own.
I want to be honest, not alarmist, because hype is the opposite of useful here. This is not a story about AI deleting your job. The same WEF report projects a net increase of around 78 million jobs by 2030. The point is narrower and more practical: the definition of a good sales or marketing professional is moving, and it is moving toward the people who run AI as a system, not the people who use it to fix typos.
There's one more number that should reassure you it isn't hype either. Despite near-universal adoption, McKinsey found only about 5.5% of organisations are seeing significant financial returns from AI, and an MIT study found roughly 95% of AI pilots show no measurable profit impact at all. The companies winning are not the ones with the most tools. They're the ones that redesigned the actual work. Exactly the same thing is true of individuals.
The tool is not the advantage. What you do with it is.
05 How to cross the line
How to actually cross the line
Here's the practical part, and I'll say up front: you do not need me, or any specific course, to do this. The source genuinely does not matter. A good book, free documentation, a YouTube channel you trust, a sharp colleague, a structured program — any of them can work. What matters is that you stop polishing tasks and start rebuilding workflows. So, concretely:
-
1
Pick one workflow you own — not a task.
Not "writing emails." A whole motion: everything from "I have a list of accounts" to "I have a booked meeting," or everything from "I have a campaign brief" to "it's live and I know if it worked." A workflow has steps. A task doesn't.
-
2
Map every step you currently do by hand — be honest about the hours.
Write them down. You'll be uncomfortable. That's the point — that discomfort is your opportunity, quantified.
-
3
For each step, change the question.
Stop asking "can AI write this for me?" Start asking "can AI do this whole step if I give it the right context?" The second question is where fluency lives.
-
4
Build the context once and reuse it.
The skill is not a clever one-off prompt. It's a repeatable input — your ICP, your voice, your rules — that you run again and again. Repeatability is the entire game.
-
5
Keep a human checkpoint where judgement actually matters.
This is how you avoid shipping workslop. AI does the volume; you own the call. That line is not negotiable, and it's also where you stay valuable.
-
6
Measure the hours you got back — then reinvest them in the human part.
The conversation. The relationship. The strategy. AI is not there to make you less human at work. Done well, it gives you room to be more.
That's it. There is no twelfth secret. Do this for one workflow and you will have crossed the line that the 88% never cross — and you'll do it on your own real work, which is the only place the learning sticks.
06 Closing
One honest closing thought
The gap I've described all the way through this piece is not a tools gap. It's not a talent gap. It's a structure gap. Almost no one was ever shown how — so almost no one went past the shallow end. That's the whole story, and the good news inside it is that anyone reading this can cross it, because the barrier was never ability.
The one thing that reliably gets people across faster is also not magic. It's structure, feedback, and doing the work on your own real projects alongside other people doing the same. That's the entire reason cohort-based learning works — not hype, just the missing scaffolding put back.
If you want that structure, my AI Accelerator is one way to get it — two tracks, one for sales and one for marketing, four weeks, built around your actual work, not toy examples. If you'd rather do it on your own from a book and a quiet weekend, genuinely, do that. I mean it. The only thing I'd ask you not to do is the thing most people do by default: stay at the shallow end because nobody ever showed you there was a deep one.
You've been told AI will change your job. Now you've been shown how. The rest is just whether you start. Browse more on the blog →
