Hi there,

I was recently struck by an ironic realization:

While we're dealing with the most advanced technology humanity has ever created, we keep falling into the most ancient of traps when it comes to using it.

And technology hasn't solved these mistakes. It has amplified them. So today, let's find out what these timeless mistakes are and understand them so we can break free.

Let's dive in!

Let's Work Together in 2025

If you're looking to implement AI profitably in your business in 2025, I'm opening a limited number of spots to work directly with me. Together, we'll uncover and implement AI opportunities that can add $10K+ in recurring profit to your bottom line.

Reply “2025” and I'll share the details.

State of AI: Record Adoption, Record Failure?

According to McKinsey, more than 70% of companies have adopted AI in at least one business function. Investment has exploded to $276.1 billion (13 times more than a decade ago). The gold rush is definitely on. But beneath all this hype, something isn't working.

While adoption is growing, it remains surprisingly shallow. Only 8% of businesses use AI in five or more functions. And more than 80% of AI projects fail.

So what's going on?

While most people blame technology complexity or talent shortage, the truth is simpler and older than AI itself. It's human behavior and organizational patterns that are killing most AI projects.

Let's explore these ancient vices and how they relate to AI adoption - starting with the most volatile one: Wrath.

Wrath (When AI Promises Meet Reality)

Ever watched a flashy AI product demo? Everything works perfectly until you actually try it in production.

Instead of getting 10x productivity, you get 10x frustration. No wonder people get angry, especially if they pay millions in license fees. A lot of AI journeys end right here: "Tried it. It didn't work."

The mistake happens when you treat AI tools like just another piece of software - something you can buy, install, and forget about.

But the reality about AI is fundamentally different. AI isn't a destination you reach - it's a journey you take, which involves continuous learning, constant adaptation, and gradual improvement. In other words: AI is only as good as the people who use it. The human factor decides AI success – not the amount of compute resources you throw at it.

That's why we shouldn't let anger and setbacks kill our AI journey. Instead, embrace the process and get your mindset right from day one.

Speaking of mindset, let's talk about the next deadly sin: Pride.

Pride ("Not Invented Here" Syndrome)

Pride in AI comes in two deadly flavors.

First, there's the "We don't need this" pride. You've probably heard it: "Our industry is different", "Our customers prefer the traditional way".

That's what Kodak said about digital cameras - the company that INVENTED the handheld camera. "People will always love film!" We all know how this story ended. Blockbuster is another example:

These companies didn't fail because they couldn't adapt. They failed because they were too proud to adapt.

Second, there's the "We can build it better" pride.

Take Bloomberg. March 2023: "We're building our own GPT! 50 billion parameters, purpose-built for finance!" One year and millions of dollars later... their custom model performs worse than publicly available ones.

This is what happens when pride takes over:

• Not-Invented-Here syndrome kicks in

• We overestimate our own capabilities

• We underestimate how complex these systems are

The question isn't whether you can build it... The question is whether you should.

Which brings us to the next sin:

Lust (The Shiny Object Syndrome)

Sooner or later, you see this "cool AI tool", and boy, do you want to have it. The pattern is always the same:

• Find an amazing AI solution

• Look for problems it might solve

• Watch the project die

When we get too excited about what AI can do, we tend to forget about what our business actually needs. Sure, playing with Generative AI is fun. Building custom GPTs is exciting. But throwing AI at random problems rarely creates value.

Instead, look for the intersection between AI capabilities and your actual business needs. No hard feelings if they don't align - move on and find something that does.

Speaking of moving (or not moving), let's talk about…

Sloth (When Comfort Kills Innovation)

Want to see what sloth looks like in the AI age? Look at the automotive industry. Tesla has a market value that's more than literally all other automakers combined. But their revenue is just a fraction of the biggest manufacturers.

What happened? It's not that everyone thinks Tesla makes the best cars. It's that the market sees Tesla as the most innovative and fast-moving company, while traditional automakers have fallen to sloth. Tesla ships Autopilot updates over-the-air while traditional carmakers are stuck (re-)organizing.

So how do we avoid sloth? I generally recommend the 20/20 rule for atomic AI projects: Small enough to implement quickly, big enough to matter, clear enough to measure success. The key is to start small, but start NOW. Because while you're analyzing, your competitors are learning.

But while some companies do too little, others try to do too much…

Gluttony (When Too Much AI Kills Success)

Imagine a starving person suddenly get access to a buffet. That's how some companies approach AI - trying to consume everything at once.

A popular example of this is McDonald's (coincidence!). They started with a seemingly simple goal: AI-powered drive-throughs. But then they tried to implement everything at once:

• Text-to-speech

• Speech recognition

• Natural language processing

• Order system integration

• Zero human backup

All of this in a complex environment with diverse accents, background noise, endless menu customizations, and (worst of all!) hungry customers. The result was a viral failure:

@themadivlog

How did I end up a butter #fyp

They could have started with a much less integrated, less automated version of this (actually great) use case. This would have reduced complexity, provided clear fallback options, and left room to learn and grow. For example, having humans take orders while AI pre-populates menu items to speed up the process.

You wouldn't eat an entire Big Mac in one bite, would you?

Which brings us to perhaps the most dangerous sin of all - greed...

Greed (Moonshots Missing The Mark)

Greed in AI isn't just about money - it's about chasing the biggest possible outcome, often at the expense of realistic gains. MORE innovation, MORE disruption, MORE market share... but at what cost?

Take Amazon's ambitious AI "just walk out” concept. A bold vision to disrupt retail entirely - no checkouts, no cashiers, just walk in and walk out. But they had to scrap it after it came out that they needed an army of (cheap) remote workers monitoring cameras to figure out what customers actually bought.

To understand why this happens, let's look at where companies choose to play:

There are essentially four playing fields. The most challenging is the top right: applying cutting-edge technology to huge problems outside your core expertise. On the bottom left, that's where you apply established technology to well-understood problems.

Amazon played the disruptor game: Operating outside their core competency (offline retail), with cutting-edge technology (computer vision to track people, items, and interactions in stores), requiring high implementation cost (hardware setups), but also offering and incredible potential upside.

Unless you have billions to spend on AI, that's not where you want to start. Most companies are better off starting in familiar problem domains and exploring use cases with relatively established AI technology before progressing to more advanced, game-changing fields.

And speaking of wanting what you can't have, let's look at our final sin - one that's making one company very, very rich...

Envy (The FOMO That's Feeding The Giants)

Believe it or not, but "envy" actually stems from the Latin word invidia. And guess who's benefiting most from everyone's AI envy?

During every gold rush, there are two kinds of people: those digging for gold and those selling shovels. But this isn't just about GPUs. It's about building technical AI capabilities in general:

• Buying Copilot licenses

• Getting ChatGPT Enterprise subscriptions

• Setting up fancy AI platforms

Now, all of these can be great - if you have a good use for them. But don't go buying shovels just because everyone around you is buying shovels.

Remember that the question isn't "What AI tools is everyone buying?", but "What problems are we trying to solve?"

Some of these problems might be a good fit for AI. But most of your business problems probably won't be.

Don't chase AI FOMO - chase profit instead.

Moving Forward

These seven sins might seem overwhelming, but understanding them gives us a clear path forward.

If you want to dig deeper into how to build successful AI initiatives, check out my previous post on 9 principles for AI success.

As my favorite philosopher once said:

The path won't be straight, but that's okay - learn and adjust.

Just make sure you're not falling for sins that are older than AI itself.

See you next Friday,
Tobias

PS: If you'd like to implement AI profitably in your business, I'm opening a limited number of spots to work directly with me to add $10K+ in recurring profit to your bottom line. Reply “2025” and I'll share the details.