Welcome to the AI implementation paradox:
Where the cure is exactly what's killing you.

It's like telling someone who's drowning, "Just swim!" (Thanks, I'm cured! 🙄)

The Steve Bartlett Advice Analysis

Full disclosure: I'm a Steve Bartlett groupie. His Diary of a CEO is my workout playlist. And his recent AI post? Pure gold.

5000+ likes, 400 reposts, and enough fire emojis to warm a small planet.

Why? Because every single recommendation is exactly what companies should do.

Also why? Because 99% of companies physically cannot do any of it.

Let's break down why his advice, while theoretically perfect, creates a practical paradox for most organizations:

1. "Treat your data like an asset class – clean it, label it and keep it in-house"

The theory: Yes! Clean data is the foundation of good AI implementation.

The reality: Who actually has clean data? Has anyone ever reached the mythical state of "our data is clean"? If you've found this data nirvana, please send your address – I'd like to make a pilgrimage.

I witnessed a large Financial Services company spend 6 months arguing about who should clean their broker data. Two years later, the data is still dirty and now called "the thorny data operations Obstacle Reduction task." People whisper about it as if it's a cursed quest.

It's the corporate version of 'I'll start going to the gym once I'm in better shape.'

The Unlock : Today's AI has the capability to clean and structure your messy data. It might not be perfect, but it can get you 70-80% of the way there – which is far better than the 0% progress you're making while waiting for data perfection.

In fact, I recently wrote a LinkedIn Post - exactly on this point: The myth of clean data

2. "Build a culture of endless small experiments"

The theory: Spot on! Experimentation is the only way to learn and adapt quickly in the AI era.

The reality: Most organizations have been engineered to do the exact opposite - perfectionism, punish failure, 50-step approval processes.

I like to call this corporate "cancel culture" – try something new that doesn't work immediately? Canceled. Suggest we skip the 17 -person approval committee? Super canceled with a side of "please go see HR."

I sat in a meeting where an executive declared "We're now a culture of experimentation" immediately followed by tasking 3 teams to build rival PowerPoints to argue for the right to run a 2-week test. Testing PowerPoint slides to prioritize tests isn't innovation, it's corporate theatre with a cover band called "innovation."

Let's be brutally honest here: true organization-wide cultural transformation takes years, not weeks or months. Building a genuine "culture of endless small experiments" across an entire company is a 2-5 year journey at minimum. But the AI train isn't waiting around – it's already pulling out of the station at exponential speed.

Do you have years to build the perfect experimental culture before getting started with AI? Or do you need results now?

The unlock? Don't fix the whole company culture at once – start with one fully empowered autonomous pod (3-5 people max) who can make decisions and actually fail without being hunted down by the "corporate immune system".

3. "Double down on EQ skills: negotiation, coaching, story-telling."

The theory: Great advice. And it’s in every leadership book published since the stone age.

The reality: Yet most formal hiring, promotion, and leadership development processes still evaluate, select, and develop talent like it’s 1999:

• Resumes or CVs screened for technical features & specs like a laptop
• Interviews focused on evaluating past accomplishments, not adaptability and potential
• Assessments that value deep domain expertise over learning agility

Real Talk: You can re-train most technical knowledge and skills every couple years. For 90% of jobs, they’re eminently learnable and repeatable.

90% of hiring managers also have no idea how to recognize or select for the most important EQ skills even for experienced leaders.

How exactly do you assess EQ or Negotiation Skills on a resume?

How do you assess EQ or Coaching Skills in a 45-min zoom interview?

The breakthrough?

• Test story-telling by having candidates explain blockchain to your grandma
• Test negotiation by dropping them into a real biz problem with competing priorities and resources
• Test adaptability by emphasizing how candidates learn over what they’ve accomplished and built
  

4. "Recalculate head-count plans and re-do your org chart assuming agents handle 60-70% of routine tasks."

The theory: Forward-thinking planning that acknowledges how AI will reshape work.

The reality: I recently worked with a 200-person company that took 18 months just to identify what their “AI use cases” i.e. their "routine tasks". Eighteen. Months.

And re: that re-org - Planning your company-wide reorg in step zero is like that 6-month old startup that took 18 months re-architecting their cloud infrastructure.

Think about all the AI waves that have hit us in the past 18 months. Think about all the babies born who are now running, talking, and jumping.

Unless, like Google, you can afford to cut 12k people overnight —and get 12k leaders committed without them quitting and torching your employee first culture— it’s basically impossible to execute a company-wide re-organization in less than 18-24 months.

The AI waves that you’re making assumptions around will likely change multiple times in that period.

The breakthrough? Re-calibrate your business value equivalent of stuck-until-breakthrough: for AI, it isn’t until you’ve already scaled and implemented it across a coalition of multiple processes stitched together.

Chose one team. Solve one pain. Tackle one process. Get one success. Let time create the organizational pull and demand, don’t push or plan it massively in advance.

Yes this approach feels like “slower”. But the alternative is doing what that 200 person did six years ago. And they’re still under 200 people today. It’s “slower” in the way doing the right thing strategically and not causing organizational paralysis is usually the fastest way to “go slow to go fast”.

5. "Automate personal drudgery (meeting booking, expense filing). Host a fortnightly 'agent demo' session."

The theory: Start small, show wins, and build momentum. Perfect!

The reality: Many companies restrict AI tool usage for legitimate security concerns.
• Regulatory requirements (4-year record keeping, bias testing)
• Governance needs (proper documentation, risk assessment)
• Trust building (transparency requirements, ethical considerations)
• Change management (proper protocols, employee concerns)

By the time legal and IT security approve tools, they're often outdated.

On the other side of this paradoxical ocean : At an AI workshop I ran a few months ago for HR leaders, they complained how terrible it is that candidates are using AI in the application process, while also explaining that they expect candidates to know how to already use AI to save the most time before they apply :). We can perform Olympic level mental gymnastics when we need to.

Meanwhile, there's this strange stigma around AI within companies. Be we either angels or techno-saints, it seems. Either pure-human or the AI savior.
Like Mrs. Maisel "I woke up like this" when everyone knows there were two hours of getting ready before.

Even Steve below in the comments on his post below wrote "(Typo in my post, I meant PHD, not PDF. At least ya know that a human wrote it 😅🤷🏽‍♂️)" - as though having an AI help write is almost a digital scarlet letter.

The unlock: Leaders need to go first. Leadership has to normalize the use of AI by actively talking about their own experiences with it, good and bad. Because trust me, your people are almost certainly using it—they're just too scared to admit.

6. "Learn to prompt: write goals that include objective, constraints and success metric in ≤3 lines."

7. "Identify your single most time-intensive process and prototype an agent-driven version within 30 days. Our CEO wants to make herself redundant with agents within 30 days. "

OK, let's take these two together, since they unlock the "ultimate AI adoption hack."

Having run hundreds of AI workshops by now, guess what the #1 concern is
EVERY. SINGLE. TIME?

"I know it's not perfect."

"I think it would be hard to learn."

Nope. It's "AI will take my job."

And here's Steve saying "Great! You SHOULD make yourself redundant!"

I even wrote a blog article about it - where AI finds out that us humans find him scary! -

And here's Steve, essentially saying "Yes! That's exactly what you should do! Make yourself redundant!"

This is basically the equivalent of telling someone how to swim by tossing them into shark-infested waters.

I've been in the room when leaders nonchalantly suggest automating entire departments, even while the heads of those departments are sitting right there. You can see their souls leave their bodies in real time. That's not a change management strategy; it's a career heart attack.

There's a reason this hack isn't more widely promoted. As Steve points out, "I'm surprised more people aren't talking about this..." Yeah, it's because they're all updating their LinkedIn statuses to 'previous role.'

The breakthrough perspective: It's not about making the PEOPLE redundant - it's making the WORK redundant.

Nobody wants to hear "we should automate your job," but you know what does get people excited?

"Let's automate the parts of your job you HATE using AI."

"What if AI could do all the boring stuff for you so you can focus on the fun stuff?"

"How could AI help you do more of what you actually ENJOY doing?

Because here's the truth - the fastest way to kill AI adoption isn't technical failure...it's turning your workforce into a bundle of existential dread. When people worry about job security they don't innovate, they don't experiment, they don't embrace new tech. They close ranks, protect, and avoid anything that rocks the boat..

The Path Forward: From Paradox to Progress

Steve’s advice isn’t bad—it’s just advice you’d need to be perfect at everything to follow.

So how do we break this cycle?
The answer isn’t to ignore it, it’s to bridge the gap from where you are today.

With:
1. One use case (and where your shoes are burning)
2. A structured process (geared towards speed e.g. AI Accelerator Lab¹)
3. A working prototype (to share with your peers)

1. Start Small, But Start Now

Don't wait for perfect data. Don't wait for cultural transformation. Don't wait for reorganization approval.

Start with one team, one use case, one problem that causes real pain. The smaller and more focused your starting point, the faster you'll see results and build momentum.

2. Use Structured Frameworks, Not Open-Ended Meetings

Traditional meetings are unsuited for solving complex challenges, brainstorming and making collaborative decisions as a group - exactly what’s needed for rapid AI transformation.

At Solved Together, we use the AI Accelerator Lab specifically to break this barrier – helping companies:

• Decide which AI use case to start with first (in 1 day)
• Design what it would look like with AI (in 2 days)
• Deliver a working AI prototype (in 2 weeks)

The results speak for themselves:

3. Build Capabilities Through Action, Not Preparation

The skills needed for the AI era – experimentation mindset, learning agility, extreme open-mindedness – can't be learned in a classroom. They're built through doing.

By starting small and showing results quickly, you create the evidence needed to justify the very cultural changes that the experts recommend as prerequisites.

Then rinse and repeat every few weeks. Once proven in one team, run in parallel across teams. That's how you scale AI adoption rapidly.

The AI train isn't just leaving the station – it's already accelerating exponentially. You can't afford to wait for your organization to transform before you begin the transformation.

As the ancient proverb says:
"The best time to plant a tree was 20 years ago.
The second best time is now."

With AI, now might be all we have left.
Unless you want to be that company still drafting your AI strategy while your competitors are drafting their "how we disrupted the industry" memoir.1

[About AI Accelerator Lab]
Our AI Accelerator Lab helps companies bridge this gap, based on a proven framework that's helped hundreds of leading enterprises, ranging from Fraunhofer, Volkswagen, SAP, and Mercedes-Benz, to Microsoft and gov. institutions all over the EU. It combines AI Design Sprint^TM and other proven neuroscience based advanced collaboration methods to help teams collaborate effectively and rapidly, specially using AI capabilities within internal transformations OR within an existing product. DM Me if you would like to learn more.