Markus Wittkowski

12 meetings in a single day aren’t proof you’re important. They’re proof your architecture is broken. Because if systems were in place to decide, delegate, and automate—your calendar wouldn’t be a battlefield.

Better questions lead to better answers… Instead of asking “How do I manage my calendar better?” ask “Why is my calendar still full—what system gap is it signaling?” — and why this is much better: because a busy calendar is not bad planning, it’s broken design.

@CernBasher This is exactly what I needed, saved me a lot of time 🙌 Appreciate you, man.

@GergelyOrosz People don’t care about matrix math when LLMs let them - write 10× faster - code without boilerplate - analyze data like a pro - brainstorm endlessly - automate entire workflows The output is the superpower. The rest is implementation detail.

Amusing how 99% of people using LLMs forget how these things work: They are advanced probability machines. They generate the next most likely token (word) based in the input and their training. Under the hood, it’s a giant matrix multiplication that has eerily good output.

@EMostaque Chat UI feels like mobile phones in 2000: clunky, familiar, not the endgame.

What is the absolute best chat ui/ux or experience for LLMs? Anyone cracked this?

@Tesla Little disappointed by Tesla’s Master Plan Part IV. All vision, no roadmap — no targets, no timelines, no concrete commitments. More manifesto than master plan. But happy they set it out anyway.

@CernBasher Really liked this piece — made me realize DCF only captures what’s visible while optionality is about what’s possible. This shift in thinking explains why most investors miss the next 100-baggers.

Most solo founders drown in tasks. The next generation doesn’t. markuswittkowski.de/design_delegat…

@alex_avoigt 💯

About Optimus As a mechanical engineer, I learned that to build tools, you need tools that are 10x more precise than what you want to build with them and since Optimus itself is a tool, you need about 10x 10=100 100 the usual precision, which is a major problem and challenge in terms of mass manufacturing. Add to that the required reliability and durability, and it becomes an enormous challenge. Finally, you have to manufacture all of this at very low cost and produce it in series. When I think about the challenges with Optimus, I underestimated the task Tesla set itself, and perhaps they underestimated it too. At the moment, Tesla may be back at the drawing board to overcome a local maximum they have encountered when it comes to the hardware and especially the hands. Tesla has to go through a lot of new inventions to achieve what they want to achieve with Optimus, so delays will be normal in the coming years and its not a quick fix. However, other companies will not develop a multifunctional humanoid robot that can perform all the tasks of a human being at all but try short cuts like they tried with autonomous driving vehicles. They will build specialised humanoid robots for just some tasks and many limitations and sell them early on, rather than developing what Tesla is trying to do. People will get the impression Tesla is falling behind but the opposite is true. In the early stages – and we are in the early stages – people will be confused and believe that other companies have overtaken Tesla, but those are the ones who do not understand that the value of a humanoid robot that can do everything a human can do is 100 or 1,000 times higher than that of a specialised humanoid robot, because you only need one for all your needs. The software is a challenge and will remain one but the hardware is the hard nut to crack as almost everything around it needs to be invented in particular to move hands with a DOFs of a human, its precicion, dexterity, sensitivity, strength and range of a human being. Right now there is only one company in the world that has the vertical integration, talent, leadership and funding to make real humanoid AI-driven robots a reality and that is Tesla. It likely will take much longer than people want to hear to achieve the goals but I have no doubt Tesla will make it happen. $tsla

AI feels powerful. Just make sure you’re behind the wheel. markuswittkowski.de/ai_drive_yours…

So AI is making intelligence cheap. Cool. But here’s the real shift—intelligence is being industrialized, just like electricity once was. That’s huge. And maybe a little scary. What happens next? markuswittkowski.de/cost_of_intell…

😂 3 reasons why programming (coding) is close to dead: - AI writes better code than you—faster, cheaper, and with fewer bugs. - It’s no longer about writing code, it’s about telling AI what you actually want. - Nobody cares how the sausage is made—they just want the app to work. RIP coding. Long live software development! ⚡

@timoreilly 🔥 After 2 years of delegating all my programming tasks to ChatGPT, I truly believe the end of programming (coding) is near. But software development will explode. We’ll design, define, and develop systems at a whole new level. But coding? Nah.

I just wrote a piece on why I think the chatter about AI putting programmers out of work is wrong. Every time it gets easier to tell computers what we want them to do, it brings more people into the fold and unleashes a wave of creativity and creative destruction. That reads like opportunity to me. oreilly.com/radar/the-end-…

Have you noticed? Code from ChatGPT-4o has become ✨visually enriched✨ lately. Log statements and comments are now sprinkled with emojis 🫠 print("🚨 Something went horribly wrong! 💥")

Mac mini and or this baby? Project Name: NVIDIA Project Digits Price: $3,000 Availability: May 2025 Hardware: - GB10 Grace Blackwell Superchip - 128 GB unified memory - Up to 4 TB NVMe storage Performance: - Supports AI models with up to 200 billion parameters - Connectable to a second system for up to 405 billion parameters Software: - Linux-based NVIDIA DGX OS - Compatible with NVIDIA AI Software Library, PyTorch, Python, Jupyter Notebooks

Big thanks to @joaomdmoura and #CrewAI for inspiring me to rethink what computers could be! 🤖💡 markuswittkowski.de/making_our_per…

Yesterday, I rewatched Star Trek: The Next Generation... and damn, where are the coders? 🤯 Is the future of tech about making coding disappear? markuswittkowski.de/no_programmers…

Love this mindset: Embracing the complexity of others challenges me to reflect on my own biases and grow from it. 🌱

LLMs are getting smarter, faster, cheaper... markuswittkowski.de/llms_four_tren…

Better questions lead to better answers... Instead of asking 'Which jobs will AI eliminate?' a more insightful question is What tasks will AI take over?