Hello there!I used to be a full time VFX artist working as compositor, 3D artist, producer and supervisor.
These days I am mostly home-bound due to health reasons, so my focus is mostly on tending to my family as the resident cook and doing the occasional remote work for film and TV projects.
You can find my work history on IMDB or LinkedIn and I do have a rather old showreel on Vimeo. Apart from that, this is my online home. Feel free to get in touch via email or on Micro.blog.
Every big empire falls sooner or later. I wonder if we are witnessing Nero’s burning of Rome right now.
It’s 30°C outside (and as much inside). Who the hell can sleep with these temperatures‽ I am seriously considering getting up and coding for a bit.
Links Update
I decided to pick up work on Links my barebones bookmark manager three days ago. Several dozen commits later it is shaping up nicely.
It now has:
- a nicer more streamlined design
- working full content search
- AI based summaries that also work for images (which is great for the search part)
- editable titles (either manually or AI generated)
- the archiver now runs as a daemon in the background
- API access so it works with iOS shortcuts, Python, etc.
- overall refactoring to be more readable code
- a randomizer that shows ten random bookmarks
- updated and expanded the documentation
Still a few things to do though like adding more file types and maybe add thumbnails for other things then images.
But it is shaping up nicely and a proper release is visible on the horizon.
I wonder. Are „proclivity“ and „pathology“ just the favorite words of Jordan Peterson or do all psychology professors talk like that?
You could easily get this episode down to half the length if you cut out those two words.
One of the things that annoys me me most about LLMs is that they ruined properly crafted typography. I love to use em dashes. I feel like it shows that I put thought & care into a text. Well, not anymore. Now it says „hey that dude used an AI to write this. Feel free to discard it as not original.“
I’ve neglected Links my bookmarking service. Too much other stuff going on like my wife’s iPad app and a new recipe app I’m trying to develop. Oh and also family life, dialysis and surgery recovery. Too much? Most likely.
📷 Micro Blog Photo Challenge Day 28: Ephemeral
Osteria Scuderia. 📍
Enjoying a nice summer day with some Spaghetti Carbonara and a glass of bitter lemon.
I really like GitKraken as my daily Git application. Check it out: Get GitKraken at 50% off
Looking for ideas
I had my cataract surgery this afternoon and I’m not allowed to read or watch anything on TV or my iPad for three days. I’m already feeling somewhat at a loss. I’m looking for ideas of what to do that won’t strain my eyes and also won’t strain my body too much, since I cannot lift heavy weights, do sports, or engage in any other strenuous activities that would increase my blood circulation or blood pressure too much.
That wasn‘t so bad. Only an hour later and I have checked off all five doctor calls.
Ah the joy of trying to get a doctors appointment in Germany on a Monday morning. Sadly, I have to reach five doctors this morning.
I finally caved and paid for a month of Claude Pro for the work on my iPad app. It should come out cheaper then using API calls for vibe coding.
Why I Embraced AI Assisted Writing for my Blog
The Old Way Wasn’t Working
I’ve always been a private blogger. No schedule. No pressure. Just me writing when inspiration strikes.
That used to be enough.
But life has a way of changing the game on you. When kidney issues landed me on dialysis, everything shifted. The energy I once had for longer writing sessions? Gone. The focus needed for those longer posts I loved crafting almost 15 years ago? Scattered.
I’d sit down with brilliant ideas bouncing around in my head. But translating them into coherent, engaging posts felt like pushing a boulder uphill. The gap between what I wanted to say and what I could actually produce became this frustrating chasm.
Writing stopped being joy. It became work.
The Unexpected Journey to AI
My path to AI-assisted writing didn’t start with writing at all.
It began with images. MidJourney caught my attention first - the idea of creating visuals from text descriptions fascinated me. I’m a coder at heart, so the concept of prompt engineering felt natural.
From there, curiosity led me to ChatGPT. Just playing around. Testing limits. Seeing what this AI thing could actually do.
TypingMind entered the picture as a better interface for managing these AI conversations. But the real breakthrough came when I discovered Anthropic’s Claude models. Something about Claude’s responses felt more… human. More collaborative.
That’s when I started thinking: if AI can help with images and casual conversations, maybe it can help with the writing that’s become so challenging.
What AI Actually Does for My Blog
Here’s what people get wrong about AI writing assistance. They think it’s about replacing the writer.
It’s not.
For me, AI bridges the gap between having ideas and having the energy to execute them properly. It’s like having a research assistant, writing partner, and editor all rolled into one - but one who never gets tired when I need to take breaks.
The AI handles the heavy lifting of research. It pulls current facts, finds relevant examples, and organizes information in ways that make sense. When writer’s block hits (and with health challenges, it hits more often), the AI helps me push through by generating different angles or approaches I hadn’t considered.
But here’s the key: I’m still driving the process. Every idea comes from me. Every direction change is my call. The AI just helps me get there without burning through the limited energy I have each day.
Since embracing this approach, I’ve published more than I have in years. Not because I have a schedule to maintain, but because writing is enjoyable again.
Keeping It Real
Let me be crystal clear about something: if a post doesn’t sound like me, it doesn’t get published.
Period.
The AI might help craft sentences or organize thoughts, but every published word has to pass my authenticity test. I work with the AI through multiple iterations, adjusting tone, refining arguments, and polishing until it genuinely feels like something I would write.
This isn’t about tricking readers or taking shortcuts. It’s about having the tools to express my actual thoughts and ideas when my body isn’t cooperating with my creative ambitions.
The iterative process is crucial. I’ll ask for revisions on specific paragraphs, request tone adjustments, or completely redirect sections that don’t feel right. The AI is patient in ways human collaborators might not be - it doesn’t get frustrated when I need the fifth version of the same paragraph.
My Current Workflow
The Setup That Changed Everything
I built my writing process around TypingMind, which lets me create specialized agents with different skills. My “Professional Blog Writer” agent is like having a writing partner who actually knows me.
The agent comes loaded with plugins that do the heavy lifting:
- Perplexity Search for current research and facts
- Sequential Thinking for structured outlines
- Memoryplugin to remember my writing style and preferences
- TubePlus for YouTube content analysis
- Firecrawl for web scraping when I need specific sources
- Image Search for visual inspiration
From Idea to Published Post
My process is surprisingly simple now. I start with just a topic idea or maybe a short summary of what’s bouncing around in my head. Nothing fancy.
The AI immediately pulls up research, creates an outline, and we go back and forth until it feels right. Sometimes that’s two exchanges. Sometimes it’s ten. Depends on the complexity and how clear my initial thoughts were.
Then comes the actual writing. Here’s where the Memory Plugin really shines - it usually nails my writing style on the first or second attempt because it remembers how I like things phrased, my sentence patterns, even my weird quirks.
The Back-and-Forth Dance
When something doesn’t feel right, I’m specific: “Make this section more conversational” or “This paragraph is too formal for my voice.” The AI adjusts, and we keep refining until I hit that moment where I think, “Yes, this sounds like me.”
The whole thing outputs in Markdown, so I just copy-paste into my blogging software. No reformatting headaches.
Why This Works for Me
The energy I used to burn on research and first drafts now goes into the creative refinement. I’m collaborating rather than grinding. And because the AI learns my style over time, each post gets easier to perfect.
Looking Forward
The future of AI-assisted blogging isn’t about replacing writers. It’s about amplifying what they are already good at while compensating for their limitations - whether those are time, energy, health, or simply the occasional creative block.
For bloggers dealing with similar challenges, my advice is simple: start small. Pick one aspect of your writing process that frustrates you most and see if AI can help with that specific piece. Don’t try to revolutionize everything at once.
The key is maintaining control. The AI is a tool, not a replacement. Your voice, your ideas, your standards - those don’t change. What changes is having the support to express them even when life throws curveballs.
I’m more excited about blogging now than I’ve been in years. Not because AI does the work for me, but because it helps me do the work I want to do, even when my body isn’t fully cooperating.
And honestly? That’s made all the difference.
This post was written with the assistance of Claude AI through TypingMind, following my usual collaborative process of research, outlining, drafting, and refinement until it properly captured my voice and experience.
From 10 Meters to 1.5 Kilometers
Eight weeks ago, I had open kidney surgery. After being bedridden for most of my three and a half week hospital stay, that final week felt like a triumph.
I’d graduated from assisted walking to making it down the hospital floor and back on my own. By discharge day, I was walking the entire floor independently. I felt confident. Ready to get back to normal life.
That confidence lasted exactly ten meters outside the hospital.
Ten meters from my front door, I needed a thirty-second break. Just ten meters. The same distance I’d been covering easily in the hospital corridors suddenly felt impossible. After feeling so capable walking those hospital floors, I felt like I’d run a marathon.
By the end of that first day home, I’d managed maybe thirty or forty meters total before admitting complete defeat. The difference between hospital walking and real-world walking hit me like a wall.
Here’s what I didn’t understand then. Hospital floors are designed for recovery. They’re flat, climate-controlled, with handrails and places to rest every few meters. The real world has curbs, uneven pavement, weather, and distances that actually matter. Walking a hospital corridor doesn’t prepare you for walking to your mailbox.
The medical reality is pretty straightforward, but no one really prepares you for the psychological whiplash. After weeks of bed rest followed by successful hospital walking, you think you’re ready. Then you discover that three and a half weeks of lying still, even with that final week of corridor walking, has fundamentally changed what your body can handle.
Those first few days home were a study in humility I never wanted to take. Each day I’d try to walk a little further. Some days I managed it. Other days I didn’t. The frustrating part wasn’t just the physical limitation—it was accepting that my hospital confidence had been completely misplaced.
I started tracking my progress. Not with meters or fancy apps, but with houses. How many houses down the street could I make it before needing a breather? Before I had to turn around? These landmarks became my measuring stick. When my confidence was shattered, making it one house further was concrete proof I was still getting better, even if it didn’t feel like it.
The breakthrough came gradually, then suddenly. Around week four post-surgery, I realized I wasn’t counting steps anymore. Week six, I was walking around the block. Week seven, I made it to the local park and back.
Yesterday, I walked 1.5 kilometers in thirty minutes without stopping once. I held a conversation the entire time. Eight weeks after surgery, four and a half weeks after that humbling ten-meter reality check.
What strikes me most isn’t just the physical progress. It’s how hospital walking creates false confidence that real-world recovery quickly corrects. Before surgery, walking was just transportation. Now it’s proof that my body works outside controlled environments. Each step feels like a small victory over both the surgery and that initial confidence crash.
The research says most open surgery patients don’t hit these milestones until week ten or twelve, especially after extended hospital stays. That feels good to hear, but it also makes me realize how misleading hospital walking can be. Your real recovery timeline starts when you leave the building.
If you’re starting your own walking recovery after major surgery and hospitalization, here’s what I wish someone had told me. Hospital walking is just the beginning. Don’t let that final week of corridor confidence fool you into thinking you’re ready for the real world. You’re not, and that’s completely normal.
Track everything once you’re home. Write down distances, times, how you felt. After that initial confidence crash, the numbers become your proof of progress when your brain tells you you’re moving backwards.
Don’t compare your home walking to your hospital walking. They’re completely different activities requiring different energy levels and skills. Hospital floors are recovery training wheels. Real-world walking is the actual test.
Consistency beats intensity every single time. I walked every day after discharge, even if it was just to the mailbox. Those tiny daily efforts compound faster than sporadic longer attempts. Your body needs to relearn movement in real environments, not just controlled hospital corridors.
Bad days will happen. Last week I had a terrible walking day where everything felt harder than usual. I questioned whether I was actually improving or just fooling myself. The next day I walked further than ever before. Recovery after that kind of confidence crash isn’t linear.
The mental side of this recovery taught me things I never expected to learn. Humility, for one. Hospital walking had made me feel capable. Real-world walking reminded me I was still healing. Both were necessary parts of the process.
It also taught me to celebrate ridiculously small victories. Making it to the corner without stopping. Walking up three stairs without getting winded. Holding a conversation while moving. After that ten-meter wake-up call, these became genuine achievements worth acknowledging.
Looking ahead, I’m still technically in the recovery window. Open surgery patients don’t hit full recovery until month three. That means I’ve got another month of potential improvement ahead of me. The thought is both exciting and surreal.
My new goal is two kilometers by week ten. Then maybe three by month three. But honestly, I’m less focused on the numbers now. The real victory is that walking feels normal again, even outside hospital corridors.
Recovery from major surgery isn’t just about getting back to where you were. It’s about developing a completely different appreciation for what your body can do in real environments, not just controlled ones. Every day I can walk 1.5 kilometers is a day I don’t take for granted.
Eight weeks ago, that confident hospital walking meant nothing when faced with ten real meters. Today, 1.5 kilometers feels routine. That’s not just physical healing—that’s rebuilding trust with your own body one step at a time, in the real world.
This post documents my personal recovery experience. Always follow your doctor’s specific instructions for post-surgery activity and recovery timelines.
Daring Fireball: Seven Replies to the Viral Apple Reasoning Paper
[…] when it comes to these systems [LLMs], I’m mostly just interested in whether they’re useful or not, and if so, how.
That feels very short sighted. If we don‘t solve the alignment problem now, it will be too late when we do get to AGIs.