Data & AI/ML Entrepreneur | Early Stage Investor

Over a decade of experience within AI/ML and Data Infrastructure as a founder, operator and investor.Passionate about bringing new and paradigm-shifting technologies to market. This passion has led me to a over decade-long entrepreneurial career focused on emerging software industries and global markets - from AI &amp; Data Infrastructure startups and venture capital funds to Fortune 500 technology leaders such as Google. Raised +$50M in early-stage funding from top VCs as a founder and operator. A big believer in people, data, design and telling a story. Passionate about building new products with the right GTM, messaging and growing teams that do so.Deep understanding of the venture capital, private equity, SaaS, cloud, data and AI ecosystems with solid global relationships at all levels.Frequent keynote speaker on topics around data, AI, startups and venture capital. Considered as one of the top AI &amp; data thought leaders in the Nordics.

Oliver Molander

The Anthropic team wrote all of Cowork (their newly launched non-technical version of Claude Code) using Claude Code 👇 

And they built it in 1.5 weeks (confirmed by Felix Rieseberg).

Holy sh*t.

Things are changing fast.

At the same time, I see founders struggling to make a simple update to their website in 1.5 weeks.

Anthropic launched an entirely new product in that time.

And as Felix noted, they’re releasing Cowork early. It has rough edges. But figuring out what to build is increasingly the hardest part of software engineering - and they believe getting feedback early and hearing what users actually need is how you build something truly good.

And this is a company valued at close to $0.5 trillion, with thousands of employees.

Founders, you have no excuse anymore.

Don’t fall into the anti-AI hype.

There’s a saying that the future is already here - it’s just not evenly distributed.

I still see plenty of developers insisting that AI coding tools are overhyped and error-prone.

But I’m also starting to see the opposite:

People who’ve fully embraced them and now find they’re so good that it’s triggering a real crisis of confidence.

If you haven’t seriously tested AI coding tools in the last year, try them again. 

Don’t assume anything.

You’ll be surprised.

#artificialintelligence

I see too many brilliant technical founders in Europe getting stuck in what I call intellectual procrastination.

They debate in their own minds what to do, weighing different outcomes - instead of just doing.

This is why high-agency founders succeed. They start doing, fail, learn, and repeat.

As Andrej Karpathy puts it:

Agency > Intelligence

Karpathy has said he intuitively had this wrong for decades, shaped by a pervasive cultural veneration of intelligence - media narratives, entertainment, obsession with IQ, etc.

Agency is significantly more powerful and significantly more scarce.

Are you hiring for agency?

Are we educating for agency?

Are you acting as if you had 10× agency?

The best thing about high agency is that it compounds.

This is something organizations like YC are especially good at: pushing founders out into action.

There’s a saying in Finnish, “Liike on lääke”, which roughly translates to “Motion is the cure.”

The best founders get anxious when there’s no motion - when nothing is happening or being tested and validated.

For too long, European cities like Stockholm - despite an incredible concentration of talent - have been held back by societal expectations: first get a master’s degree, then gain experience at a big company, and only then maybe think about building something. 

But the truth is, you don’t need permission - you can just do things.

The tides are turning fast, though.

Now is the time of the high agency builder.

#artificialintelligence

One of the biggest myths right now that needs to be busted is that Claude Code is only for coding.

It can be used for so much more.

As my friend Antoine Moyroud recently said, we need to stop treating AI like a chatbot.

The gap between power users of AI and casual chat users is going to widen very fast. Start experimenting with Claude Code now.

Non-technical people shouldn’t be afraid of the CLI.

You can already leverage Claude Code for almost any kind of knowledge work:

Sales, marketing, growth, fundraising, law, taxes - you can become surprisingly competent quickly.

It’s hard to overstate how big the gap will be between people who understand how massive this shift is and those who don’t.

As Greg Isenberg put it:

You’re just a few .md files away from outperforming 99% of people on this planet.

2026 is the year Claude Code grows beyond coding.

#artificialintelligence

Probably not a biggie, but the creator of Node.js just admitted that the era of humans writing code is over.

#artificialintelligence

Here's a shout-out to all young technical operators from the Nordics working at the frontier in leading AI companies - you embody the future.

Gabriel Petersson 
Erik Kaunismäki 
Jonatan von Martens 
emil fagerholm 
Emil Ahlbäck 
Minette Kaunismäki 
Christian Ryan 
Matheus Mendes 
Eric Zakariasson
Jonathan Granskog

...and many more that I've missed.

The future is in the youth.

#artificialintelligence

Vincent Granville

Chief AI Scientist, GenAItechLab.com

Full-stack entrepreneur with multimillion-dollar exit. Co-Founder at BondingAI.io. Author (Wiley, Elsevier). Creator of open-source Python libraries / Web APIs such as xLLM and GenAI-Evaluation. Builder of high performance in-memory agentic multi-LLMs for professional users and enterprise, with real-time finetuning, self-tuning, no weight, no training, no latency, no hallucinations, no GPU. Made from scratch, leading to replicable results, leveraging explainable AI, adopted by Fortune 100. With a focus on delivering concise, exhaustive, relevant, and in-depth search results, references, and links. 

Vincent is a leading expert, mathematician, patent owner, former post-doc at Cambridge University, former VC-funded executive, with 20+ years of corporate experience including CNET, NBC, Visa, Wells Fargo, Microsoft, eBay.  Vincent also founded and co-founded a few startups, including one with a successful exit (Data Science Central acquired by Tech Target). He is also a top influencer for brands such as Nvidia, SingleStore, and The Linux Foundation.

From 2005 to this day, Vincent successfully arbitraged the real estate market, buying, selling and renting out homes with significant growth and return over time: Building wealth starting from negative net worth when he first arrived in the US in 1995.

Chief AI Scientist

Privately funded GenAI and ML Optimization research lab, creating and disseminating open source intellectual property. Developing explainable and generative AI, synthetic data, computer vision, NLP and generic ML technologies. Check out our certifications, Python libraries, eBooks, articles, courses and podcasts, including our best seller "Synthetic Data and Generative  AI" (Elsevier). Visit our company website at MLtechniques.com.

I’m happy to share that I’m starting a new position as CAIO at BondingAI!

Chief AI Officer. Building architecture, IP, and new AI agents for xLLM, including: • Predictive analytics on retrieved tables from PDFs, databases, the Internet or Excel repositories • Anomaly detection for fraud detection / cybersecurity • Medical data: automatically recovering insights (electrocardiogram data, predicting heart attacks) and data compression • Tabular data synthesis: best and fastest on the market, no deep neural network involved • No-Blackbox deep neural networks and other alternatives (96% correct prediction for next token)

See also new book on these topics, entitled "No-Blackbox, Secure, Efficient AI and LLM Solutions", at https://lnkd.in/g5cmxHE3

See also my alternative to DNNs, where there is no gradient descent, and the minimum of the loss function is obtained in closed form with no iterative algorithm. There is a whole chapter on the topic in my new book, see table of contents at https://lnkd.in/g5cmxHE3

Quantitative Researcher | Author of O’Reilly’s Deep Learning for Finance

Most people default to MSE when training regression models.

That’s fine, until it isn’t.

Mean Squared Error (MSE) aggressively penalizes large errors.
- Great when your data is clean.
- Terrible when it isn’t.

In real financial and economic data, outliers are not bugs, they are features.

That’s where Huber loss quietly does a better job.

What’s the difference?
• MSE squares every error
 → Large mistakes dominate the optimization
 → Models become overly sensitive to rare shocks

• Huber loss behaves differently
 → Quadratic for small errors (like MSE)
 → Linear for large errors (like MAE)

In plain terms:
Huber protects your model from being hijacked by a few extreme observations.

Source: LinkedIn

"If solved, it would destroy modern cryptography."
No, it would not. Modern cryptography does not rely on prime numbers. You get a lot better with irrational numbers. Details in my book 0 and 1 – From Elemental Math to Quantum AI at https://lnkd.in/gQD86nAP

An MIT lecturer just used ChatGPT to "understand" a problem mathematicians haven’t solved in 160 years.

And he failed big time.

Let me explain what actually happened.

He published a piece on Forbes showing his ChatGPT conversations on the Riemann Hypothesis.

It's one of the deepest unsolved problems in mathematics. 
If solved, it could destroy modern cryptography.

And there's a 1 million $ prize for anyone who is capable of doing so. 

The lecturer claimed he didn’t understand it, as he is not a mathematician. But he believed ChatGPT could help.

He tried to prove this by asking very naive and wrong questions to ChatGPT, just like a layman would do.

And in the end, he did not understand the problem.
He just understood ChatGPT’s simplifications of his own confusion.

This is the new failure mode we’re not talking about enough.

1) Generative AI doesn’t increase understanding, it just increases confidence.

ChatGPT is very good at: 
– sounding coherent
– regurgitating internet materials
– producing explanations that feel complete, but are usually wrong

What it CANNOT do is guarantee correctness, depth, novelty, or epistemic understanding.

But the user doesn’t see that. They see fluency.

That’s textbook Dunning–Kruger, now automated.

2) Explaining is different from knowing.

The article proudly shows how GPT can: 
– simplify definitions
– paraphrase known explanations
– restate standard intuitions and basic concepts

There is no mathematical insight or reasoning. Just random words.

The Riemann Hypothesis isn’t unsolved because people don’t understand the math. 

It’s unsolved because existing demonstration attempts failed.

No amount of conversational fluency changes that.

3) AI doesn’t know when it’s wrong, and neither do non-expert users.

ChatGPT doesn’t know:
– which simplifications are harmless
– which nuances are essential
– which mistakes invalidate the reasoning

And users who lack domain expertise cannot tell the difference.

So errors don’t look like errors. They look like insights.
That’s the danger.

The article concludes that ChatGPT helps us understand why the Hypothesis is unsolved.

But the interaction itself shows the opposite: 
– shallow questions
– circular explanations
– analogies that replace rigor
– philosophical confusion dressed as insights

This is just an illusion of understanding.

It's not true that AI lowered the barrier to knowledge. What it did was lowering the barrier to "feeling informed".

And that’s far more dangerous.

Because when everyone believes they understand everything, expertise collapses, standards erode, and confidence detaches from competence even more.

And that’s exactly why automation bias + Dunning-Kruger effect is such a dangerous mix.

Fluency is not understanding. 
And AI is very good at masking the difference.

----
Follow me Chiara Gallese, Ph.D. for more on AI & Tech risks

taplio

The Battle

Oliver Molander

Vincent Granville

Followers

Posts

Engagements

Likes

Performance ScoreOliver Molander's LinkedIn Performance

Performance ScoreVincent Granville's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Oliver Molander

Vincent Granville

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Oliver Molander's LinkedIn Performance

Performance ScoreVincent Granville's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreOliver Molander's LinkedIn Performance