For the last two years, “doing something with AI” has become the most expensive non-strategy in business.

In boardrooms across industries, leaders feel the pressure. Competitors are announcing pilots. Headlines are screaming about trillion-dollar valuations. Consultants are selling urgency. And yet, beneath the noise, very few companies can point to measurable impact on the balance sheet.

In this Offscript conversation, we sat down with Sebastian - former Porsche Digital and Mercedes-Benz digital leader, founder, and current CEO of Y1 - to talk about what actually works when companies try to turn technology into value.

Spoiler: it has very little to do with building your own LLM.

Digital Transformation Was Never About Technology

One of the most uncomfortable truths Sebastian shared is this:

Digital transformation is not a technology problem. It’s an organizational one.

For over a decade, enterprises have tried to “transform” by hiring agencies, launching one-off digital products, or bolting innovation labs onto legacy structures. Most failed, not because the tech didn’t work, but because incentives, ownership, and accountability never changed.

Ironically, the much-maligned “silo” isn’t the enemy. In many cases, it’s the only structure that actually works at scale. The mistake is assuming you need to tear everything down, instead of building new capabilities alongside existing ones, then integrating them once they’re proven.

Mercedes-Benz didn’t become more digital by forcing transformation everywhere. They created a separate capability, ran it like a startup, and injected it back once it worked.

That distinction matters.

AI Fails When You Start With Tools Instead of Purpose

Most AI initiatives collapse for one simple reason:

They start with what instead of why.

“We need something with an LLM” is not a strategy.

“We want to reduce customer churn by 15%” is.

Sebastian made this painfully clear: companies that mandate AI usage without a clear business objective end up building impressive demos that solve nothing. Meanwhile, many real opportunities don’t require generative AI at all — just automation, better data pipelines, or classical machine learning models that have existed for decades.

The irony? Some of the most impactful AI use cases aren’t flashy enough to get budget approval - until someone reframes them in business terms.

Machine Learning ≠ Generative AI (And That Confusion Is Costly)

One of the most valuable parts of the conversation was dismantling the idea that LLMs are the answer to everything.

They’re not.

Many of the hardest, highest-value problems, fraud detection, portfolio optimization, marketing efficiency, pricing models, are machine learning problems, not generative ones. LLMs are interfaces. Powerful ones. But they’re terrible at multi-dimensional optimization.

Treating them as universal intelligence leads to expensive misapplications.

The companies that win won’t be the ones building the biggest models — they’ll be the ones orchestrating the right tools for the right problems.

Hiring for the AI Era: Curiosity Beats Credentials

When asked how he spots A-players early, Sebastian didn’t mention degrees, resumes, or years of experience.

He mentioned:

• Curiosity

• Impatience

• Willingness to challenge ideas

People who ask “why does this work?” and “can this be better?” adapt faster than those who wait for instructions. In fast-moving environments, skill gaps can be trained. Stagnation can’t.

That mindset becomes even more critical as tools evolve faster than job descriptions.

Agents, Automation & The Future of Work

Will jobs disappear? Some will. That’s not new.

The more interesting question is whether organizations use AI to eliminate people — or to eliminate friction. The companies seeing real impact aren’t replacing humans with agents. They’re reclaiming time, reducing waste, and letting people focus on higher-leverage work.

Agent-based systems matter here. Not as chatbots, but as task-specific components that can be swapped, upgraded, or outsourced without rebuilding everything.

Think systems, not silver bullets.

Final Thought: Don’t Be Late, But Don’t Get Drunk Either

Sebastian’s closing line sums it up perfectly:

“Don’t be late to the party - but don’t drink too much.”

AI isn’t optional anymore. But neither is discipline.

The winners won’t be the loudest adopters. They’ll be the ones who understand their business deeply enough to know where technology actually belongs.

WATCH THE EPISODE HERE:

About Penomo

Penomo is a digital asset infrastructure platform specializing in tokenized energy and AI infrastructure financing.* Through tokenization technology, Penomo is streamlining financing processes, enhancing liquidity, and enabling efficient financing for the global energy transition and AI expansion.

The time to lead is now!

→ Curious if tokenized capital fits your project? Take 2 minutes to find out.

Connect with us: Website | X | Telegram | TG Announcements | Discord | LinkedIn