Introduction
Konnichiwa! Welcome to the AI Automation Dojo. Today, someone handed us a beautiful thirty-six-page report titled “AI Adoption in the Enterprise”, and we read all of it so you don’t have to – which is lucky, because it turns out to be a horror novel with a sales brochure stapled to the back.
I’m your host, Andrzej Kinastowski, one of the founders of Office Samurai, where we read other people’s research the way you eat a fish: keep the meat, spit out the bones, and never buy the boat from the guy who caught it.
So, whether you’re a leader losing sleep over a strategy you can’t quite admit isn’t working, or an employee quietly pasting the company financials into a free chatbot on the bus home, you are in the right place.
Now grab your favorite katana (or a highlighter for the parts of the report they were hoping you’d skip), and let’s get to it!
The WRITER 2026 Report: A horror story in PDF
Imagine someone hands you a thirty-six-page report.
It’s glossy. It’s got those expensive purple gradients on the cover. The kind of PDF that clearly cost more than my first car. And across the top, in big confident letters, it says: “AI Adoption in the Enterprise”.
So you start reading. And it is thirty-four straight pages of human suffering. CEOs lying awake at three in the morning. Employees quietly committing sabotage. Interns being out-prompted by the new guy. Robots getting promoted to senior management. It reads less like a business report and more like a Stephen King novel where the monster is a Microsoft Copilot license.
And then you get to page thirty-five. And the report – having walked you slowly, lovingly, through the entire burning building – turns to you, dusts off its hands, smiles, and says: “By the way! Companies that use OUR specific product see a 333% return on investment. Care to schedule a demo?”.
Today we are discussing the WRITER 2026 AI Adoption Survey. And I want to be fair to WRITER, because this is actually a good piece of research. They partnered with a firm called Workplace Intelligence, they surveyed twelve hundred C-suite executives and twelve hundred regular employees, across about thirty industries, in the US, the UK and Ireland, France, the Benelux countries, and Germany. This was fielded at the end of last year and the start of this one. So it’s fresh. It’s big. It’s real data.
It’s also, and I cannot stress this enough, an ad. WRITER sells an enterprise AI platform. Which means this entire report is a company saying “the building is on fire”, while standing in the doorway, holding a fire extinguisher, with a price tag on it.
But that’s fine. We’ve done this before. Remember the McKinsey and MIT reports from last year, you read these reports the same way you eat a fish. You keep the meat. You spit out the bones. And whatever you do, you do not buy the boat from the guy who caught it.
And here’s the bone I want you to keep in your mouth the whole episode, because it’s the most important detail in the document, and they bury it in the methodology section where they know nobody reads: every single person they surveyed already uses AI, at a company that already permits it. These are not the stragglers. These are not the dinosaurs. This is the cool kids’ table. This is the data from the people who are supposedly winning.
So when I tell you it’s a horror show – just remember, this is the highlight reel.
Let’s get into what’s not working. Because, oh boy.
Why AI is a “Massive Disappointment”
Let’s start with the headline number, the one that should make every Chief AI Officer spit out their oat-milk flat white.
Nearly half of executives – 48% – admit that AI adoption at their company has been, and I’m quoting the report directly here, a “massive disappointment”.
A massive disappointment. Not “a learning journey”. Not “early innings”. Not “we’re iterating toward value”. A massive disappointment. That is the kind of language you use about a blind date gone wrong, not a multi-year, multi-million-dollar enterprise technology strategy.
And the money backs it up. Only twenty-nine percent of them say they’ve seen significant return on investment from generative AI. For AI agents, it drops to twenty-three percent. So roughly three out of four executives have spent real money, on real licenses, with real consultants, and gotten, financially speaking, a warm feeling and a slide deck.
Long-time listeners will recognize this. We have a name for it. It’s the AI Paradox. Everybody’s flying the warp-speed spaceship to the end of the driveway to pick up the mail. Different report, same driveway.
Now here’s where it goes from “disappointing” to “what are we doing as a species”. Because you’d think, faced with no return on investment, a sensible company would slow down. Maybe figure out a plan. Maybe ask, “what are we actually trying to achieve here?”
Instead: sixty-nine percent of companies are doing layoffs because of AI. Sixty-nine percent. They are cutting human beings, today, on the promise of AI productivity.
And in the very same report – thirty-nine percent of them admit they don’t actually have a formal strategy to make any money from that AI. Let me put those two facts in the same sentence, because the report politely keeps them on separate pages, like divorced parents at their kid’s wedding. Two-thirds of these companies are confident enough in their AI to fire people. But almost four in ten of them have no plan for how the AI generates revenue. They are absolutely certain about the part where humans leave, and completely fuzzy on the part where value arrives.
That’s not a strategy. That’s a hostage situation with a vision board.
And it gets better, because seventy-five percent of them – three out of four – admit that their company’s AI strategy is, again quoting, “more for show than for actual internal guidance”. Three-quarters of executives are looking you dead in the eye and confessing the strategy is a Halloween costume. It’s a LinkedIn post wearing a trench coat. It exists to be seen having a strategy, which is a very different activity from having a strategy.
This, by the way, is the entire reason Office Samurai exists. You know our strict nobullshit rule. These companies have looked at that principle, and decided to do the exact opposite, at scale, while firing the witnesses.
Now – and I want you to sit with this one – it is not getting better. It’s getting worse. The survey ran this same study last year, so we can watch the trend line. And the trend line is going the wrong way down a one-way street.
The share of executives who say AI is “tearing my company apart” went from forty-two percent last year to fifty-four percent this year. The share saying it’s created “power struggles and disruption” went from forty-two to fifty-six. And satisfaction with their own AI tools didn’t just stall – it dropped on every single dimension they measured. The big one: the number of people who rate their tools “excellent” on security and data governance fell seventeen points in twelve months, from seventy-one percent down to fifty-four.
Think about that. They’ve had a whole extra year of practice. A year of conferences. A year of webinars. A year of guys like me talking about how to do it. And they got worse at it. This is the only field of human endeavor where experience appears to be a disadvantage. It’s like watching someone take golf lessons for a year and come back unable to hold the club.
So how’s the C-suite holding up, emotionally, through all this? Funny you should ask. Seventy-three percent of CEOs say their company’s AI strategy is causing them stress or anxiety.
Thirty-eight percent say the stress is high or crippling stress. Crippling. That’s the word.
Sixty-one percent of executives fear they could personally lose their job if they fail to lead their company through this AI transition. So they’re firing other people because of AI, while privately terrified that AI is going to get them fired. It’s the circle of life, but the lion is also crying.
And here is my favorite sentence in the entire report, the one I’d put on a billboard. Fifty-eight percent of these executives admit that many of their fellow executives lack the fundamental knowledge to make strategic decisions about AI.
Read that again. The people in the room have looked around the room, and the majority verdict is: “the other people in this room have no idea what they’re doing”. Which would be a hilarious thing to discover, if it weren’t the same room that just approved the layoffs.
And then we arrive at the cherry on top. Seventy-five percent of these executives expect AI agents to be part of their company’s C-suite – the actual leadership team – within the next five years. They want to give the robot a seat at the big table.
In the same report, thirty-five percent of them admit they are not confident they could “pull the plug” on a rogue AI agent if it started causing damage.
So let me make sure I have the plan straight. We’re promoting it to leadership. We’re handing it real authority. And one in three of us isn’t sure we could turn it off if it went bad. That’s not a digital transformation. That’s the first ten minutes of every AI apocalypse movie.
Alright. Take a breath. That was the executive suite. Now let’s go downstairs, to the people who actually do the work. Because if you’re a manager listening to this – and you are, that’s who listens to this show – this next part is about you. And it’s not going to be comfortable.
Employee sabotage and the Manager trust gap
Managers and directors. This chapter’s for you. Sit up straight.
The report asked employees a simple question: is your manager an AI champion? Someone who actually encourages the team to embrace this stuff?
Thirty-five percent said yes.
Thirty-five. So roughly two out of three of your people do not see you as a champion of the single most disruptive technology of their careers. The most common answer, at fifty-eight percent, was that the manager is “open to AI but provides minimal direction or encouragement”, which is corporate for “he forwarded us a TED talk and went back to his calendar”. And then there’s a brave seven percent who say their manager flat-out doesn’t like or understand AI. Somewhere out there, a manager is listening to this in his car, nodding along going “yeah, those guys are the worst,” with absolutely no idea his own team put him in the seven percent.
But that’s the warm-up. Here’s the gut-punch. Seventy-five percent of employees say they would trust AI more than their own manager for at least one work task.
Among Gen Z, it’s eighty percent.
And it’s not for trivial stuff. Forty-one percent would rather have AI than their manager analyze their performance data. Twenty-two percent would rather get feedback on their work from the machine than from you. And fifty-five percent of employees – sixty-four percent of the young ones – believe they personally know more about using AI for their job than their manager does.
So just to summarize the relationship: your team thinks they understand the tools better than you, they’d rather the algorithm did your performance reviews, and a solid chunk of them would prefer career advice from a chatbot that, few episodes ago, enthusiastically approved bringing a parrot into our office.
Now, you could get defensive about that. Or you could notice that this is the single biggest open goal in modern management. Because the bar is on the floor. Thirty-five percent. If you become even a vaguely competent AI champion for your team, you are instantly in the top third. You don’t have to be a genius. You just have to show up and not be the seven percent.
Let’s keep going, because it gets spicier. The sabotage.
Twenty-nine percent of employees admit they are actively sabotaging their own company’s AI strategy. Among Gen Z, it’s forty-four percent – nearly half.
And I want to be clear here: kudos to GenZ. Sylwia Królikowska always says that GenZ has the best bullshit radar. And since we have just learned that 75% of executives find that their AI strategy is “more for show than for actual internal guidance” (which is just corpo speak for bullshit), maybe the sabotage is the appropriate reaction? And I need you to hear how they’re doing it, because the report lists the methods, and it reads like a union playbook for the machine age.
They’re using unapproved tools. They’re refusing to use the AI outputs. They’re ignoring the guidelines. Some are intentionally generating low-quality outputs to make the AI look bad. And my personal favorite – they are tampering with performance metrics so that the AI appears to be underperforming.
That is not laziness. That is a coordinated misinformation campaign run by your own staff against a software product. That is quiet quitting that went to night school and learned prompt injection. And seventy-five percent of the C-suite say employee sabotage is a serious threat to the company’s future, so they know it’s happening. They just can’t prove which of the smiling faces in the Monday stand-up is the one feeding garbage to the model. And why are they doing it? The report asked.
Number one reason, at thirty percent: “I don’t want AI to take over my job”. Then, twenty-eight percent: “AI has too many security issues”. Twenty-six percent: “my company’s AI strategy is poorly executed”. Another twenty-six: “AI diminished my value or creativity”.
Look at that list. Only one of those is really about fear. The rest are product reviews. Your employees aren’t all Luddites clutching their staplers. A lot of them are quality control, and the feedback is: this rollout is bad and you should feel bad.
And finally, the thing I think is the real story of this whole report: the two-tiered workplace.
Ninety-two percent of the C-suite say they are actively cultivating a new class of “AI elite” employees. Their words – “AI elite”. And sixty percent say they plan to lay off employees who can’t, or won’t, use AI.
So a new aristocracy is forming inside your company, in real time. And it pays. The report’s “AI super-users” are about three times more likely than the laggards to have gotten both a promotion and a raise in the past year. They’re saving themselves nearly nine hours a week. And – this is the part that should terrify every middle manager who came up the traditional way – they skew young. Forty-three percent of Gen Z are super-users, versus twenty-five percent of the Boomers.
So the org chart is quietly being redrawn. Not by tenure. Not by who’s been loyal for fifteen years. By who’s best friends with a chatbot. The kid who joined eighteen months ago and treats Claude like a colleague is on the escalator. The veteran who “prefers to do it himself” is on the list. That’s the comfort zone, ladies and gentlemen, and it is on fire, and the people holding the marshmallows are twenty-six years old.
Alright. I’ve been beating you up for fifteen minutes. Let me throw you a rope. Because there is good news in here, and it’s actually useful.
The solution: Human-in-the-loop and hard data
Here’s the thing the doom merchants leave out: when AI works, it really works.
In this same report, eighty-six percent of employees and ninety-seven percent of executives say they’ve benefited from AI in at least one concrete way. Employees are saving around six hours a week. The super-users we just talked about – nearly nine. Executives, around twelve. That’s not nothing. That’s a full extra workday, every single week, for a big chunk of the workforce. The value is real. The spaceship can fly. People are just pointing it at the mailbox.
So what separates the companies getting twelve hours back from the forty-eight percent calling it a “massive disappointment”? Conveniently, the report tells us, in its little “how to fix it” section near the end. And I read that section, and I laughed out loud, because it is – almost word for word – the stuff we’ve been saying on this show for thirty episodes.
Their fix number one: tie AI to measurable business outcomes. Define what success looks like before you build, then actually track it. That’s every one of our Process Discovery episodes. That’s the whole “ROI of sanity” matrix – business value on one axis, technical feasibility on the other. Don’t automate vibes. Automate something you can put a number on and show your CFO.
Their fix number two: empower your business users to innovate, and stop forcing everything through an IT bottleneck. That’s the Center of Excellence – what we called the “Adulting Phase”. Govern enough that it’s not the Wild West, but not so much that everyone quits and goes back to the speakeasy. It’s a balancing act, and the data we just covered shows you exactly what happens when you get it wrong in either direction.
Their fix number three: invest in growth, not just cost-cutting. In other words – stop using the warp-drive to fire people and start using it to do things you literally could not do before. The companies treating this purely as a headcount-reduction tool are the ones doing layoffs with no revenue strategy. Funny how that connects.
And their fix number four, and this is the big one: governance for AI agents, with – and I want to frame this and hang it on the wall – a human who can actually pull the plug. Remember, thirty-five percent of executives weren’t confident they could stop a rogue agent. The fix for that isn’t a longer ethics committee meeting. It’s the principle Office samurai hammers on every single time we talk about agents: Human-in-the-Loop. The AI does the grueling work, the AI drafts the answer, the AI preps the booking and a human being clicks “approve”. You want the agent to prepare the package. You do not want it to have its finger on the button, unless you are genuinely relaxed about a language model interpreting a rounding error as a reason to dissolve the board.
If you don’t put a human and a guardrail between GenAI’s enthusiasm and your business, you don’t get transformation. You get a disaster.
Now, the report also has a fifth point, which is – and I’ll be honest with you, because that’s the rule here, #NoBullshit – their point is “work with a supportive technology partner” and then page thirty-five reveals that the supportive technology partner they had in mind is, astonishingly, themselves. So take that one with the appropriate pinch of salt
But here’s the truth I’ll give you for free, even though it does, admittedly, pay my mortgage: the MIT data we covered a couple of episodes back found companies that bring in an external partner hit roughly double the success rate of the ones trying to build it all alone. And the reason isn’t magic. It’s that an outsider doesn’t care whether Bob in accounting likes robots. An outsider will look at your favorite, beloved, decade-old process and say the thing nobody internal is allowed to say: “this shouldn’t be automated. It should be deleted”. Internal teams get stuck in the politics. We bring the katana. That’s the whole job.
So that’s the recipe. Tie it to outcomes. Govern without strangling. Keep a human on the kill switch. Fix the process before you automate it. None of it is glamorous. None of it will get you a keynote slot. It just happens to be the difference between twelve hours a week saved per employee and a “massive disappointment”.
Foundations over technology
So what did we learn today, other than the fact that you should never give a corner office to something you’re not confident you can switch off?
We learned that the most honest document about the state of enterprise AI is, ironically, a sales brochure. Thirty-four pages of carnage and one demo button. And both parts are true at the same time. That’s the thing to hold onto. The carnage is real – the disappointment, the sabotage, the crying CEOs, the robot they want to promote but can’t switch off. And the upside is also real – the twelve hours a week saved, the thirteen hundred CVs screened in forty-eight hours instead of two weeks, the super-users running circles around everyone.
The difference between those two outcomes was never the technology. It was never about having the best model or the most licenses. It was the boring stuff. Strategy before layoffs. Process before automation. A human with their finger on a kill switch.
And here’s my one ask, the single thing I want you to take away. Read the reports. All of them. McKinsey, MIT, WRITER, whoever publishes the next glossy one. Keep the data – it’s genuinely good. But when you hit the page where the building’s on fire and they’re holding the only extinguisher with a price tag on it, just smile, and remember that the fire and the extinguisher came from the same box.
Now if you’ve spent this whole episode in a slowly rising panic, because you’ve just realized your company’s entire AI governance strategy is a Slack channel called “Cool AI Stuff” and a lot of optimism – don’t panic. That’s literally what we do. Before you fire anyone, before you buy anything, before you give the robot a corner office, you find out what’s actually worth fixing. Go check out our past episodes – we’ve mapped the whole thing out: the Process Discovery, the playbook, the ladder you climb one rung at a time without breaking your corporate neck. Go have a look. I promise it’s more useful than whatever thought-leadership webinar you’re currently ignoring on your second monitor.
