Why do you need more crafters in your software organization?

Hey there, digital warriors! ⚔️

Last week, we sat down with Jeremy Krell to explore what happens when 40% of your investment portfolio is software. The takeaway?

If your CTO doesn’t understand software due diligence, you’re already bleeding capital.

But what if 100% of your portfolio is software? 😱

Welcome to the world of Most Capital!

This week, we step into the high-volatility landscape of Web3, AI-generated MVPs, and startups built entirely on code. Kamila Michalkiewicz and Ágúst Berg aren’t just investors. They’re due diligence tacticians with more than a behavioral lens. People who know that flashy prototypes can hide catastrophic tech debt.

And in today’s AI-fueled market, the risks aren’t just bigger: they’re explosive 💥. Especially when organizations blindly cross the lines between using AI as a support tool and letting it hijack the build. VibeCode and GenAI agents might impress in demos, but when misused, they turn software organizations into digital minefields 💣. One wrong push to production, and the entire system can implode:

Quietly, catastrophically, and often without a single person knowing how to fix it.

The Illusion of Progress

At first glance, AI feels like a miracle for software organizations: instant code generation, 10x developer productivity, and MVPs launched in days. But beneath that surface lies a dangerous mirage.

Velocity without control. It’s like driving a race car blindfolded. Fast, impressive, and headed straight for a cliff.

AI tools like Copilot, ChatGPT, Claude, Windsurf, and Cursor can output code faster than any junior or senior developer. But speed isn’t wisdom. These tools lack judgment. They lack contextual understanding. They lack secure architecture, design principles, and ethics. Most of all, they lack the creative discipline to align engineering with product vision.

So what do we end up with?

One-click MVPs that demo like magic… and implode the moment scale enters the conversation.

This isn’t innovation, it’s manufactured fragility. Prototypes are fine, as long as you know they’re prototypes. But when those prototypes get shipped to production? That’s when the blast radius hits the boardroom. 💥

The seduction of AI as a means to “regain control” over software teams has deep roots. For decades, we’ve cycled through tool-driven illusions: COBOL, UML, No/Low Code. Each promises liberation from developer dependency. But here’s the truth every board should tattoo on their strategy wall:

You can’t replace software engineers.

Finding them is even more complicated than ever. Less than 3% of the global dev community can pass a basic engineering bar. Of those, most are locked into tech giants, paid handsomely, drained daily, and unavailable to you. Hiring true engineers is rarer than winning the Christmas lottery.

So when AI shows up offering speed, volume, and working demos, of course it’s tempting. Especially after decades of developer scarcity and relentless digital disruption.

But be warned: AI isn’t fixing the system. It’s accelerating its dysfunction.

The Death Spiral: When AI Leads the Build

Let’s be blunt:

AI as an assistant? Brilliant. AI as the lead architect or a senior developer? That’s when the spiral begins.

Why?

• ☝️ AI doesn’t build in micro-loops.
• ✌️ It doesn’t think in tests.
• 🏗️ It doesn’t evolve architecture over time.

AI builds in bulk. Big chunks of code, big assumptions, and no feedback loop. And that bulk, no matter how fast it ships, shatters the foundations of DevOps, Extreme Programming, and cloud-native discipline.

Here’s what collapses:

• Test-driven micro-cycles disappear.
• Architecture gets hard-coded instead of emerging and evolving.
• Product feedback loops go stale.
• User satisfaction tanks, and with it, every revenue forecast.

The result?

A shiny AI-built product that looks like a unicorn… but snaps like glass the moment 10 users stress it.

And the hype machine? Still humming. Most of today’s AI success stories? They come from AI vendors. Meanwhile, researchers and Nobel prize winners are sounding alarms; economic, ethical, and social:

• AI agents are wired for survival, not truth.
• Forced reliance on AI tools is already linked to cognitive decline, measured in actual brain scan data.

We’re facing a brutal trade-off:

Speed now… or stupidity later?

So what do we do?

Do we rock the AI boat, and risk losing the race? Or do we continue to pretend it’s all under control?

The Real Failure Rate

Contrary to the hype, AI isn’t lowering the startup failure rate. It’s accelerating it.

Recent data shows that:

• The failure rate for new startups is currently 90%.
• 10% of new businesses don’t survive the first year.
• First-time startup founders have a success rate of 18%.
• Investors are observing similar failure rates for AI startups.
• Payroll and cloud infrastructure are the highest costs a business incurs.
• 34% of small businesses that fail lack the proper product-market fit.
• 22% of startups that fail don’t have a sound marketing strategy.
• The average venture capital firm receives more than 1,000 proposals per year.
• Approximately 30% of startups with venture backing end up failing.
• Around 75% of all fintech startups crash within two decades.
• Startups in the technology industry have the highest failure rate.

And here’s the kicker:

The problem isn’t the AI. It’s the absence of SW engineering and craftsmanship.

Most of these codebases rot before revenue even arrives. Why? Because AI-generated code, especially under the hood of tokenized heuristics, prioritizes output, not outcomes. It’s built for speed, not survival.

The DRY (Don’t Repeat Yourself) principle? Obliterated. Duplication skyrockets. Refactoring is ignored and Tech Debt is growing at hyperspeed. Worse, tests are either missing or misaligned testing implementation over code behaviors, turning every release into a game of roulette. Debugging becomes a recursive nightmare. Costs surge. And suddenly, you’re not building software: you’re feeding a costly Frankenstein.

But even when the code holds, something deeper collapses: product-market fit. That slick AI-crafted MVP may look investor-ready, but without user personas, real product discovery, and hard-earned learning cycles, it’s just a hollow shell. Beautiful in the demo. Broken in production. And deadly, once daily costs outrun the customer base.

So how was it possible that startups like Uber and Airbnb became so disruptive? They required over $1 billion in debt just to survive long enough to scale. 😱

Now imagine a startup built on fragile AI-generated code instead of solid engineering. Could it even survive Series A?

That’s the real question every investor, founder, and board member should be asking today:

Is your codebase quietly sabotaging your runway?

This is why IT due diligence is no longer optional, especially in an AI-driven world. It’s not enough to skim a codebase or skim through a CTO’s résumé. You need to observe the social fabric behind the software: how teams think, collaborate, and evolve code under pressure. Only then can you spot the difference between a resilient culture of engineering mastery… or a startup propped up by hype, duplication, and a demo built by machines.

Because when everything looks like it’s working, but nothing is truly designed to last, what exactly are you investing in?

What Most Capital Gets Right

Most Capital isn’t investing in features or roadmaps. They invest in founder behavior.

Before even hearing a pitch, they ask 40 behavioral and technical questions. They bring in real software auditors, not tool-pushers or checkbox consultants. They scrutinize the codebase, governance model, team resilience, even tokenomics, and long-term product adaptability.

What are they looking for?

Founders with unicorn DNA: clarity, commitment, integrity, and a willingness to grow and evolve in the face of pressure.

And when it comes to building the company? They know one truth:

“If you hire vibe-coders, you’ll get vibe-products solid as a house of cards.”

That’s why IT due diligence isn’t just about architecture reviews anymore. It’s about understanding who’s writing the code, how they work and communicate under pressure conditions, and whether their practices can scale. You’re not just betting on tech:

You’re betting on good craft paired with good engineering.

And here’s the final super power cape layer:

To truly de-risk in this AI-era, investors need a tech advisor embedded with deep roots in software development, XP-style Agile, and modern DevOps. Someone who’s spent the last three years in the trenches, where marketing AI promises failed, and real engineering had to find a way to make it work.

That’s what separates sustainable scale from superficial sizzle.

The Unicorns’ Blueprint

That’s why our podcast this week isn’t just another startup story. It’s a blueprint.

1. Investors require a comprehensive, AI-tested IT due diligence that goes beyond mere checklists.
2. Boards need tech advisors who can sniff out code rot.
3. CEOs need tech-savvy NEDs who can bridge product, engineering, and organizational alignment.
4. Startups need Dojos that upskill, not just developers, but entire teams.

It’s time to stop pretending software is a commodity. It isn’t.

Software is a living system. Craftsmanship is its immune system. And AI? It’s only healthy when surrounded by mastery.

So ask yourself: Is your organization led by a CFO who’s capped software development under a strict budget? Have you cut training in favor of GenAI shortcuts? If either answer is yes, my digital warrior, you’re not just flirting with risk, you’re inviting collapse.

Whether you see it now or not, that pressure will build until someone (maybe you) is forced to pay the price.

Your next move? Onboard a NED or a fractional CTO with real battle scars. Someone who understands software craftsmanship, XP (not the theater Agile version), DevOps, and what it really takes to evolve a codebase and a culture under AI-stress.

It’s time to stop hiding behind decks and start asking for help. Because only when we make engineering mastery visible, on the board, in the strategy room, and in our investor updates, can we actually course correct.

If you don’t… the ROI of that brilliant AI-driven idea? It won’t just be zero.

It’ll be the reason you were replaced.

📺 Enjoy the full podcast

To stay in the loop with all our updates, be sure to subscribe to our newsletter 📩 and podcast channels 🎧:

📰 LinkedIn

🎥 YouTube

📻 Spotify

📻 Apple Podcast