Marc Beierschoder

The guy, who explains why organisations struggle to adapt to accelerating technology.

[Marc Beierschoder @MBeierschoder]

Before giving an agent tools, specify its permissions, spending limits, escalation triggers and audit trail.

Then test failures, not just successes.

The new control question is: what may the system DO without asking? #AgenticAI

[Marc Beierschoder @MBeierschoder]

Your AI strategy is probably governed as a technology program.

But when software starts approving payments, changing records and contacting customers, you're not just deploying technology.

You're delegating authority. A thread.

[Marc Beierschoder @MBeierschoder]

Imagine an agent that spots a vendor discrepancy.

Can it flag it? Hold a payment? Change supplier details? Contact the vendor?

Those are four different levels of authority. A generic 'human in the loop' policy doesn't define them.

[Marc Beierschoder @MBeierschoder]

Your company might spend millions using AI to make 200 approval steps 30% faster.

And never ask why they exist.

That's an expensive way to preserve an outdated operating model.

Before automating a workflow, ask who would miss it if it disappeared.

[Marc Beierschoder @MBeierschoder]

This doesn't mean large firms lose. They still have capital, trust, relationships and distribution.

But it means 'we are bigger' may stop being an answer to 'why will we win?'

Which of your advantages survives if coordination gets cheap? #Strategy

[Marc Beierschoder @MBeierschoder]

Big companies spent decades turning size into an advantage.

AI raises a more uncomfortable question: what happens when smaller companies can coordinate like giants? A thread.

[Marc Beierschoder @MBeierschoder]

Now imagine a smaller competitor with agents handling routine reconciliation, reporting, routing and internal handoffs.

It doesn't need to replicate your management structure to challenge your execution speed.

[Marc Beierschoder @MBeierschoder]

Europe could win the AI model race. And still lose the AI economy.

@arthurmensch, CEO of Mistral AI, argues that Europe must build its own frontier models to avoid technological dependence.

I respect his ambition. But I challenge the premise.

𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝗔𝗜 𝘄𝗮𝗿 𝗺𝗮𝘆 𝗻𝗼𝘁 𝗯𝗲 𝗮𝗯𝗼𝘂𝘁 𝗺𝗼𝗱𝗲𝗹𝘀 𝗮𝘁 𝗮𝗹𝗹.

Models are becoming increasingly interchangeable. Value is shifting toward orchestration, enterprise data, business processes and the decisions AI can actually execute.

Two companies. Same AI model.

One builds a chatbot.

The other eliminates half the steps in a critical business process.

Who wins?

Meanwhile, Europe debates sovereign models while struggling with expensive energy, fragmented markets and excessive regulation.

And what if today’s LLM architecture turns out to be transitional?

We risk spending billions catching up in a race whose rules may change.

Sovereignty matters. But sovereignty without competitiveness is an expensive illusion.

𝗘𝘂𝗿𝗼𝗽𝗲 𝗱𝗼𝗲𝘀𝗻’𝘁 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗼𝘄𝗻 𝗲𝘃𝗲𝗿𝘆 𝗹𝗮𝘆𝗲𝗿 𝗼𝗳 𝗔𝗜.

𝗜𝘁 𝗻𝗲𝗲𝗱𝘀 𝘁𝗼 𝗰𝗮𝗽𝘁𝘂𝗿𝗲 𝗺𝗼𝗿𝗲 𝗼𝗳 𝗶𝘁𝘀 𝗲𝗰𝗼𝗻𝗼𝗺𝗶𝗰 𝘃𝗮𝗹𝘂𝗲.

Otherwise, we’ll end up with sovereign models and an economy dependent on everyone else’s innovation.

#AI #Europe

𝘐𝘯𝘴𝘱𝘪𝘳𝘦𝘥 𝘣𝘺 𝘈𝘳𝘵𝘩𝘶𝘳 𝘔𝘦𝘯𝘴𝘤𝘩’𝘴 𝘢𝘳𝘵𝘪𝘤𝘭𝘦 “𝘉𝘶𝘪𝘭𝘥 𝘰𝘳 𝘴𝘶𝘳𝘳𝘦𝘯𝘥𝘦𝘳.”

[Marc Beierschoder @MBeierschoder]

I once watched someone struggle through a workshop and dismissed their thinking.

Then came a brilliant follow-up email.

I had confused presentation with judgment. AI makes that mistake even easier: polished output can hide weak thinking. #Leadership

[Marc Beierschoder @MBeierschoder]

𝗔𝗜 𝘀𝗵𝗼𝘂𝗹𝗱 𝗺𝗮𝗸𝗲 𝗰𝗼𝗿𝗽𝗼𝗿𝗮𝘁𝗲 𝗵𝗲𝗮𝗱𝗾𝘂𝗮𝗿𝘁𝗲𝗿𝘀 𝗹𝗲𝘀𝘀 𝗽𝗼𝘄𝗲𝗿𝗳𝘂𝗹. 𝗪𝗲’𝗿𝗲 𝘂𝘀𝗶𝗻𝗴 𝗶𝘁 𝘁𝗼 𝗱𝗼 𝘁𝗵𝗲 𝗼𝗽𝗽𝗼𝘀𝗶𝘁𝗲.

Watch this extraordinary artwork. A human figure breaks free from its frame.

I wish more companies would let their people do the same.

Having worked at Accenture, IBM and Deloitte, I’ve seen how differently global organizations operate.

One thing I’ve learned: global scale and local autonomy aren’t opposites. When designed well, they reinforce each other.

Yet in many corporations, someone sitting thousands of miles away has more authority than the person actually facing the customer.

Another global policy. Another approval. Another decision made without understanding the local reality.

And now we’re adding AI.

𝗔𝗿𝗲 𝘄𝗲 𝗮𝗯𝗼𝘂𝘁 𝘁𝗼 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗲 𝘁𝗵𝗲 𝗯𝘂𝗿𝗲𝗮𝘂𝗰𝗿𝗮𝗰𝘆 𝘄𝗲 𝘀𝗵𝗼𝘂𝗹𝗱 𝗯𝗲 𝗲𝗹𝗶𝗺𝗶𝗻𝗮𝘁𝗶𝗻𝗴?

In 1945, Friedrich Hayek explained why central planning struggles.

Knowledge is scattered across people. Much of it is local, constantly changing and never documented.

No central authority can capture it all.

Eighty years later, we risk using AI to concentrate even more decisions at the centre.

But imagine the opposite.

AI empowers local teams to understand markets, assess risks and make decisions that previously required three layers of approval.

Shared technology. Common standards. Strong governance.

But decisions made where the knowledge actually exists.

𝗛𝗲𝗿𝗲’𝘀 𝘁𝗵𝗲 𝗶𝗿𝗼𝗻𝘆.

We’re spending billions teaching machines to make decisions while refusing to trust our own people to make them.

𝗦𝘁𝗼𝗽 𝘂𝘀𝗶𝗻𝗴 𝗔𝗜 𝘁𝗼 𝗿𝗲𝗶𝗻𝗳𝗼𝗿𝗰𝗲 𝘁𝗵𝗲 𝗳𝗿𝗮𝗺𝗲.

𝗨𝘀𝗲 𝗶𝘁 𝘁𝗼 𝗹𝗲𝘁 𝗽𝗲𝗼𝗽𝗹𝗲 𝗯𝗿𝗲𝗮𝗸 𝗼𝘂𝘁.

𝘐𝘯𝘴𝘱𝘪𝘳𝘦𝘥 𝘣𝘺 𝘍𝘳𝘪𝘦𝘥𝘳𝘪𝘤𝘩 𝘏𝘢𝘺𝘦𝘬’𝘴 𝘛𝘩𝘦 𝘜𝘴𝘦 𝘰𝘧 𝘒𝘯𝘰𝘸𝘭𝘦𝘥𝘨𝘦 𝘪𝘯 𝘚𝘰𝘤𝘪𝘦𝘵𝘺 𝘢𝘯𝘥 𝘖𝘭𝘪𝘷𝘦𝘳 𝘡𝘪𝘮𝘮𝘦𝘳’𝘴 𝘸𝘳𝘪𝘵𝘪𝘯𝘨.

𝘈𝘳𝘵 𝘤𝘳𝘦𝘥𝘪𝘵: 𝘛𝘢𝘯𝘺𝘢 𝘎𝘰𝘮𝘦𝘭𝘴𝘬𝘢𝘺𝘢 (@tanya)

#ArtificialIntelligence #Leadership #Transformation

[Marc Beierschoder @MBeierschoder]

The answer isn't unlimited AI spending. It's smaller funding tranches, explicit stop rules and decisions based on evidence—not annual promises.

If AI learns monthly while capital moves annually, which one is really setting your strategy? #AITransformation

[Marc Beierschoder @MBeierschoder]

Your AI teams are learning in weeks. Your investment committee still funds them once a year.

This may be the most expensive mismatch in your AI strategy. A thread.

[Marc Beierschoder @MBeierschoder]

Picture a team discovering a breakthrough workflow in February. They need new integration capacity in March. The budget window opens in November.

By then, another team may have built a workaround—or a competitor may have moved.

[Marc Beierschoder @MBeierschoder]

@bundeskanzler

[Marc Beierschoder @MBeierschoder]

@chams_builds So true!

[Marc Beierschoder @MBeierschoder]

Your next AI strategy may be constrained by a data-centre decision made years ago.

Boards discuss models and agents. Far fewer ask where those systems can run, at what cost and under whose control.

That's a strategic blind spot. A thread. #AIInfrastructure

[Marc Beierschoder @MBeierschoder]

@i_am_fabs Na sicher 🤡

[Marc Beierschoder @MBeierschoder]

𝗔𝗺𝗲𝗿𝗶𝗰𝗮 𝗶𝘀 𝘁𝗿𝘆𝗶𝗻𝗴 𝘁𝗼 𝗺𝗮𝗸𝗲 𝗰𝗮𝗿𝘀 𝗳𝗹𝘆. 𝗘𝘂𝗿𝗼𝗽𝗲 𝗶𝘀 𝗯𝗿𝗶𝗻𝗴𝗶𝗻𝗴 𝗯𝗮𝗰𝗸 𝟭𝟵𝟰𝟴.

One is experimenting with the limits of engineering. The other is reinventing nostalgia.

Different products, different purposes. But the contrast made me think about something much bigger.

Imagine a 50-year-old European man.

Married. Two children. Successful career.

Yet he still lives in his father’s house.

He pays most of the bills. But his father makes the decisions.

Who gets a key to the house. How the family spends its savings. Which risks are acceptable. Whether the old heating system should be replaced.

When things go wrong, his father explains why every decision was necessary.

His mother reassures everyone that there was never a better alternative.

Meanwhile, the bills keep rising. The house deteriorates. The children wonder what kind of future awaits them.

The son complains endlessly.

𝗕𝘂𝘁 𝗵𝗲 𝗻𝗲𝘃𝗲𝗿 𝗺𝗼𝘃𝗲𝘀 𝗼𝘂𝘁.

Because leaving means making decisions. Taking risks. Accepting failure. And losing the luxury of blaming someone else.

Does this sound familiar?

After decades working with European companies on technology and transformation, I’ve seen a similar contradiction.

We celebrate innovation. Until someone proposes something genuinely disruptive.

Then come the committees, approvals, risk assessments and requests for guarantees.

Everyone wants progress. Few want to own the consequences.

Nearly 200 years ago, Alexis de Tocqueville warned about democracies that could keep their citizens in a permanent state of childhood.

𝗣𝗲𝗿𝗵𝗮𝗽𝘀 𝗘𝘂𝗿𝗼𝗽𝗲’𝘀 𝗿𝗲𝗮𝗹 𝗽𝗿𝗼𝗯𝗹𝗲𝗺 𝗶𝘀𝗻’𝘁 𝘁𝗵𝗮𝘁 𝗼𝘂𝗿 𝗹𝗲𝗮𝗱𝗲𝗿𝘀 𝘁𝗿𝗲𝗮𝘁 𝘂𝘀 𝗹𝗶𝗸𝗲 𝗰𝗵𝗶𝗹𝗱𝗿𝗲𝗻.

𝗜𝘁’𝘀 𝘁𝗵𝗮𝘁 𝘄𝗲’𝘃𝗲 𝗯𝗲𝗰𝗼𝗺𝗲 𝗰𝗼𝗺𝗳𝗼𝗿𝘁𝗮𝗯𝗹𝗲 𝗯𝗲𝗶𝗻𝗴 𝗼𝗻𝗲.

We demand freedom, then ask governments to protect us from every consequence of exercising it.

And perhaps that’s why the future increasingly feels like something happening elsewhere.

𝗪𝗵𝗮𝘁 𝗶𝗳 𝗘𝘂𝗿𝗼𝗽𝗲 𝗱𝗼𝗲𝘀𝗻’𝘁 𝗻𝗲𝗲𝗱 𝗺𝗼𝗿𝗲 𝗽𝗿𝗼𝘁𝗲𝗰𝘁𝗶𝗼𝗻, 𝗯𝘂𝘁 𝗺𝗼𝗿𝗲 𝗮𝗱𝘂𝗹𝘁𝘀 𝘄𝗶𝗹𝗹𝗶𝗻𝗴 𝘁𝗼 𝘁𝗮𝗸𝗲 𝗿𝗲𝘀𝗽𝗼𝗻𝘀𝗶𝗯𝗶𝗹𝗶𝘁𝘆?

#Europe #Innovation

𝘝𝘪𝘥𝘦𝘰 𝘦𝘥𝘪𝘵: @𝘚𝘱𝘳𝘦𝘦𝘣𝘢𝘣𝘺𝘭𝘰𝘯 (𝘟). 𝘍𝘰𝘰𝘵𝘢𝘨𝘦: 𝘛𝘦𝘴𝘭𝘢 𝘙𝘰𝘢𝘥𝘴𝘵𝘦𝘳 & 𝘊𝘪𝘵𝘳𝘰𝘦̈𝘯 𝟤𝘊𝘝.