desunit
@desunit.bsky.social
Entrepreneur
✨ http://rwiz.ai - Handling reviews with AI
🎹 http://pianocompanion.info - Chords dictionary app with 1M+ downloads.
🕹️ http://chordiq.info - Learn chords.
📝 desunit.com - my blog
✨ http://rwiz.ai - Handling reviews with AI
🎹 http://pianocompanion.info - Chords dictionary app with 1M+ downloads.
🕹️ http://chordiq.info - Learn chords.
📝 desunit.com - my blog
About a week ago, in the comments under one post, I saw the usual take:
AI is dumb. It can’t solve new math. It’s not creative. It never will be.
Now let’s look at reality.
AI is dumb. It can’t solve new math. It’s not creative. It never will be.
Now let’s look at reality.
February 11, 2026 at 7:11 PM
About a week ago, in the comments under one post, I saw the usual take:
AI is dumb. It can’t solve new math. It’s not creative. It never will be.
Now let’s look at reality.
AI is dumb. It can’t solve new math. It’s not creative. It never will be.
Now let’s look at reality.
❗90% of humanoid robots sold last year came from China 🇨🇳
It’s the same EV playbook:
> Darwinian market -strongest products survive
> early state backing
> dense local supply chains
> ship chaos/fix later (I really believe that this is the best way to build products)
It’s the same EV playbook:
> Darwinian market -strongest products survive
> early state backing
> dense local supply chains
> ship chaos/fix later (I really believe that this is the best way to build products)
February 9, 2026 at 7:12 PM
❗90% of humanoid robots sold last year came from China 🇨🇳
It’s the same EV playbook:
> Darwinian market -strongest products survive
> early state backing
> dense local supply chains
> ship chaos/fix later (I really believe that this is the best way to build products)
It’s the same EV playbook:
> Darwinian market -strongest products survive
> early state backing
> dense local supply chains
> ship chaos/fix later (I really believe that this is the best way to build products)
The interface was the moat... until recently.
Found quite interesting ideas in the article: The Crumbling Workflow Moat: Aggregation Theory’s Final Chapter
Found quite interesting ideas in the article: The Crumbling Workflow Moat: Aggregation Theory’s Final Chapter
February 6, 2026 at 7:16 PM
The interface was the moat... until recently.
Found quite interesting ideas in the article: The Crumbling Workflow Moat: Aggregation Theory’s Final Chapter
Found quite interesting ideas in the article: The Crumbling Workflow Moat: Aggregation Theory’s Final Chapter
I shipped a custom CRM in a couple of days. For my wife.
She runs a home management company, and issuing invoices in Excel was a very tedious and error-prone task. One house could easily take a couple of days to align and collect everything.
She runs a home management company, and issuing invoices in Excel was a very tedious and error-prone task. One house could easily take a couple of days to align and collect everything.
February 4, 2026 at 7:18 PM
I shipped a custom CRM in a couple of days. For my wife.
She runs a home management company, and issuing invoices in Excel was a very tedious and error-prone task. One house could easily take a couple of days to align and collect everything.
She runs a home management company, and issuing invoices in Excel was a very tedious and error-prone task. One house could easily take a couple of days to align and collect everything.
Software development as we’ve known it for decades is over.
I read a great post that perfectly captures what many of us in software already feel.
The author starts from the famous Linus Torvalds quote:
Talk is cheap. Show me the code.
I read a great post that perfectly captures what many of us in software already feel.
The author starts from the famous Linus Torvalds quote:
Talk is cheap. Show me the code.
February 2, 2026 at 7:13 PM
Software development as we’ve known it for decades is over.
I read a great post that perfectly captures what many of us in software already feel.
The author starts from the famous Linus Torvalds quote:
Talk is cheap. Show me the code.
I read a great post that perfectly captures what many of us in software already feel.
The author starts from the famous Linus Torvalds quote:
Talk is cheap. Show me the code.
From what I can see around me, this is already happening. Being a junior dev today is quite tough - AI can already do a lot of entry-level work better and faster.
January 30, 2026 at 7:16 PM
From what I can see around me, this is already happening. Being a junior dev today is quite tough - AI can already do a lot of entry-level work better and faster.
I love Claude - no matter what people say.
...aaand that Reddit thread only proves the feeling - according to the discussion, even Microsoft employees are happily using Claude internally while officially selling Copilot to the outside world.
...aaand that Reddit thread only proves the feeling - according to the discussion, even Microsoft employees are happily using Claude internally while officially selling Copilot to the outside world.
January 28, 2026 at 7:18 PM
I love Claude - no matter what people say.
...aaand that Reddit thread only proves the feeling - according to the discussion, even Microsoft employees are happily using Claude internally while officially selling Copilot to the outside world.
...aaand that Reddit thread only proves the feeling - according to the discussion, even Microsoft employees are happily using Claude internally while officially selling Copilot to the outside world.
The 2025 AI Index Report is out (published by Stanford Institute for Human-Centered Artificial Intelligence).
January 26, 2026 at 7:17 PM
The 2025 AI Index Report is out (published by Stanford Institute for Human-Centered Artificial Intelligence).
About a couple of years ago, my app ChordIQ was removed by Google because it didn’t comply with the new terms.
The subscription wasn’t clearly saying it was a price per year - it only said "PRO for a year". That was enough to get it taken down.
The subscription wasn’t clearly saying it was a price per year - it only said "PRO for a year". That was enough to get it taken down.
January 25, 2026 at 7:16 PM
About a couple of years ago, my app ChordIQ was removed by Google because it didn’t comply with the new terms.
The subscription wasn’t clearly saying it was a price per year - it only said "PRO for a year". That was enough to get it taken down.
The subscription wasn’t clearly saying it was a price per year - it only said "PRO for a year". That was enough to get it taken down.
Fix this gene.
(almost whispering)
Please, oh please… don’t mess this up.
Let’s imagine a world where you talk to ChatGPT or Claude and say something like that.
Last year, a baby with a single broken gene would have died.
(almost whispering)
Please, oh please… don’t mess this up.
Let’s imagine a world where you talk to ChatGPT or Claude and say something like that.
Last year, a baby with a single broken gene would have died.
January 23, 2026 at 7:17 PM
Fix this gene.
(almost whispering)
Please, oh please… don’t mess this up.
Let’s imagine a world where you talk to ChatGPT or Claude and say something like that.
Last year, a baby with a single broken gene would have died.
(almost whispering)
Please, oh please… don’t mess this up.
Let’s imagine a world where you talk to ChatGPT or Claude and say something like that.
Last year, a baby with a single broken gene would have died.
Space race 2.0 is here
Low Earth Orbit is already packed because of thousands of satellites, more launches every month, growing collision risks and more space junk ... and now China is talking about deploying up to 200,000 satellites (currently Starlink has ~9,4k satellites). 🧵👇
Low Earth Orbit is already packed because of thousands of satellites, more launches every month, growing collision risks and more space junk ... and now China is talking about deploying up to 200,000 satellites (currently Starlink has ~9,4k satellites). 🧵👇
January 21, 2026 at 7:15 PM
Space race 2.0 is here
Low Earth Orbit is already packed because of thousands of satellites, more launches every month, growing collision risks and more space junk ... and now China is talking about deploying up to 200,000 satellites (currently Starlink has ~9,4k satellites). 🧵👇
Low Earth Orbit is already packed because of thousands of satellites, more launches every month, growing collision risks and more space junk ... and now China is talking about deploying up to 200,000 satellites (currently Starlink has ~9,4k satellites). 🧵👇
The human bottleneck problem
The more I work with AI, the more I realize humans are the real constraint; it's not execution but orchestration. 🧵👇
The more I work with AI, the more I realize humans are the real constraint; it's not execution but orchestration. 🧵👇
January 19, 2026 at 7:19 PM
The human bottleneck problem
The more I work with AI, the more I realize humans are the real constraint; it's not execution but orchestration. 🧵👇
The more I work with AI, the more I realize humans are the real constraint; it's not execution but orchestration. 🧵👇
Everyone’s using the same AI models now. So why do some teams still win?
Because context is the real moat.
Same GPT. Same Claude. Same Gemini but .... at the same time getting different results.
One team gets a generic bot.
Another gets something that feels custom-built for their product. 🧵👇
Because context is the real moat.
Same GPT. Same Claude. Same Gemini but .... at the same time getting different results.
One team gets a generic bot.
Another gets something that feels custom-built for their product. 🧵👇
January 16, 2026 at 7:17 PM
Everyone’s using the same AI models now. So why do some teams still win?
Because context is the real moat.
Same GPT. Same Claude. Same Gemini but .... at the same time getting different results.
One team gets a generic bot.
Another gets something that feels custom-built for their product. 🧵👇
Because context is the real moat.
Same GPT. Same Claude. Same Gemini but .... at the same time getting different results.
One team gets a generic bot.
Another gets something that feels custom-built for their product. 🧵👇
Remember when I wrote about Stack Overflow’s death? Well, here’s another take - and it shows that nothing is purely black or white. There’s a huge 50-shades-of-grey area in between.
‼️ Revenue doubled to $115M BECAUSE of AI ‼️
‼️ Revenue doubled to $115M BECAUSE of AI ‼️
January 14, 2026 at 7:13 PM
Remember when I wrote about Stack Overflow’s death? Well, here’s another take - and it shows that nothing is purely black or white. There’s a huge 50-shades-of-grey area in between.
‼️ Revenue doubled to $115M BECAUSE of AI ‼️
‼️ Revenue doubled to $115M BECAUSE of AI ‼️
Wrote a new post about why I choose curiosity over fear.
How random ideas turned into real projects just by following my interests and showing up every day. Inspired by Paul Graham’s fantastic post How to Do Great Work.
How random ideas turned into real projects just by following my interests and showing up every day. Inspired by Paul Graham’s fantastic post How to Do Great Work.
January 13, 2026 at 2:18 AM
Wrote a new post about why I choose curiosity over fear.
How random ideas turned into real projects just by following my interests and showing up every day. Inspired by Paul Graham’s fantastic post How to Do Great Work.
How random ideas turned into real projects just by following my interests and showing up every day. Inspired by Paul Graham’s fantastic post How to Do Great Work.
Can you imagine the chaos quantum computing could bring?
Everything tied to public-key encryption would be at risk - Bitcoin/SSL/TLS/VPN/SSH… the list keeps going. 🧵👇
Everything tied to public-key encryption would be at risk - Bitcoin/SSL/TLS/VPN/SSH… the list keeps going. 🧵👇
January 9, 2026 at 7:17 PM
Can you imagine the chaos quantum computing could bring?
Everything tied to public-key encryption would be at risk - Bitcoin/SSL/TLS/VPN/SSH… the list keeps going. 🧵👇
Everything tied to public-key encryption would be at risk - Bitcoin/SSL/TLS/VPN/SSH… the list keeps going. 🧵👇
In 2025, solo founders now make up 36.3% of all new startups - the highest share in over 50 years (source: Carta). It’s been steadily climbing since 2019.
Thanks to AI, one person can now build, ship, market, and support a product from start to finish. No huge team is required. 🧵👇
Thanks to AI, one person can now build, ship, market, and support a product from start to finish. No huge team is required. 🧵👇
January 7, 2026 at 7:13 PM
In 2025, solo founders now make up 36.3% of all new startups - the highest share in over 50 years (source: Carta). It’s been steadily climbing since 2019.
Thanks to AI, one person can now build, ship, market, and support a product from start to finish. No huge team is required. 🧵👇
Thanks to AI, one person can now build, ship, market, and support a product from start to finish. No huge team is required. 🧵👇
2025 was the year code became cheap.
With agentic tools, writing code is no longer the hard part but directing work is. The problem is that shipping didn’t get 5x-10x faster. You still need:
🧠 transforming a problem into the minimum tech solution
👀 reviewing outcomes, not lines of code
With agentic tools, writing code is no longer the hard part but directing work is. The problem is that shipping didn’t get 5x-10x faster. You still need:
🧠 transforming a problem into the minimum tech solution
👀 reviewing outcomes, not lines of code
January 5, 2026 at 7:19 PM
2025 was the year code became cheap.
With agentic tools, writing code is no longer the hard part but directing work is. The problem is that shipping didn’t get 5x-10x faster. You still need:
🧠 transforming a problem into the minimum tech solution
👀 reviewing outcomes, not lines of code
With agentic tools, writing code is no longer the hard part but directing work is. The problem is that shipping didn’t get 5x-10x faster. You still need:
🧠 transforming a problem into the minimum tech solution
👀 reviewing outcomes, not lines of code
This is the number of questions asked on Stack Overflow. Take a look at 2024 and 2025.
Yes - it drops to nearly ZERO.
This single image shows a tectonic shift in how development works today.
There are three possible explanations: 🧵👇
Yes - it drops to nearly ZERO.
This single image shows a tectonic shift in how development works today.
There are three possible explanations: 🧵👇
January 4, 2026 at 7:19 PM
This is the number of questions asked on Stack Overflow. Take a look at 2024 and 2025.
Yes - it drops to nearly ZERO.
This single image shows a tectonic shift in how development works today.
There are three possible explanations: 🧵👇
Yes - it drops to nearly ZERO.
This single image shows a tectonic shift in how development works today.
There are three possible explanations: 🧵👇
I’ve already shared this before, but AI music just keeps getting better and better. Stumbled on the track from "Ash and Echo" in my Discover Weekly and I have to admit, it sounds really good.
January 2, 2026 at 7:16 PM
I’ve already shared this before, but AI music just keeps getting better and better. Stumbled on the track from "Ash and Echo" in my Discover Weekly and I have to admit, it sounds really good.
Happy New Year! 2025 was productive AF. 2026, I hope you bring even more wins and fewer dumb mistakes - the harder the path, the stronger we get! 🎉🔥
January 1, 2026 at 4:05 PM
Happy New Year! 2025 was productive AF. 2026, I hope you bring even more wins and fewer dumb mistakes - the harder the path, the stronger we get! 🎉🔥
Google was at risk of losing its dominance - until it promoted Josh Woodward. This story is another reminder of how much one individual actually matters.
Not a process
Not an org chart
A person
Someone who sets a north star, removes obstacles, and then steps back. 🧵👇
Not a process
Not an org chart
A person
Someone who sets a north star, removes obstacles, and then steps back. 🧵👇
December 31, 2025 at 7:15 PM
Google was at risk of losing its dominance - until it promoted Josh Woodward. This story is another reminder of how much one individual actually matters.
Not a process
Not an org chart
A person
Someone who sets a north star, removes obstacles, and then steps back. 🧵👇
Not a process
Not an org chart
A person
Someone who sets a north star, removes obstacles, and then steps back. 🧵👇
You like it or not, vibe coding is now part of development. And metrics will eventually be used to understand how well each developer works with LLMs.
It’s not that different from algorithms:
> There’s O(1) - cost barely grows 🧵👇
It’s not that different from algorithms:
> There’s O(1) - cost barely grows 🧵👇
December 29, 2025 at 7:11 PM
You like it or not, vibe coding is now part of development. And metrics will eventually be used to understand how well each developer works with LLMs.
It’s not that different from algorithms:
> There’s O(1) - cost barely grows 🧵👇
It’s not that different from algorithms:
> There’s O(1) - cost barely grows 🧵👇
I’ve seen this play out over and over - real AI progress never starts with a mandate. It starts with someone being curious/staying late/breaking things/experimenting/quietly fixing a real problem/etc.
December 28, 2025 at 7:12 PM
I’ve seen this play out over and over - real AI progress never starts with a mandate. It starts with someone being curious/staying late/breaking things/experimenting/quietly fixing a real problem/etc.
Cursor dropped AI visual editing - same old story that designers should code because the end product of our interfaces is code.
Should they? Is it close to programming vs coding theme too? 🧵👇
Should they? Is it close to programming vs coding theme too? 🧵👇
December 26, 2025 at 7:12 PM
Cursor dropped AI visual editing - same old story that designers should code because the end product of our interfaces is code.
Should they? Is it close to programming vs coding theme too? 🧵👇
Should they? Is it close to programming vs coding theme too? 🧵👇