Senko Rašić
@senko.net
I help startups with AI, tech, product, and open source strategy.
Ex Pythagora (YC W24), MusicBox (acq.), AWW (acq.), Naslovnica (acq.), WebCampZg.
Personal: https://senko.net
Work: https://senkorasic.com
Ex Pythagora (YC W24), MusicBox (acq.), AWW (acq.), Naslovnica (acq.), WebCampZg.
Personal: https://senko.net
Work: https://senkorasic.com
From a squirrel:
"We interview devs by giving them a task to climb a tree. What is surprising: 14/15 rabbits failed this screening."
"We interview devs by giving them a task to climb a tree. What is surprising: 14/15 rabbits failed this screening."
From a CTO at a startup:
"We interview devs by giving them a task to build an app on the spot, from scratch (2x BE endpoints, some frontend.) They can use AI, ofc - and we dig into why they did this or that.
What is surprising: 14/15 devs from Meta failed this screening."
"We interview devs by giving them a task to build an app on the spot, from scratch (2x BE endpoints, some frontend.) They can use AI, ofc - and we dig into why they did this or that.
What is surprising: 14/15 devs from Meta failed this screening."
November 15, 2025 at 12:06 PM
From a squirrel:
"We interview devs by giving them a task to climb a tree. What is surprising: 14/15 rabbits failed this screening."
"We interview devs by giving them a task to climb a tree. What is surprising: 14/15 rabbits failed this screening."
Reposted by Senko Rašić
Zet CLI aplikacija. Mislim da samo mene veseli, ali neka. :)
ZET TUI demo
Recorded by ihabunek
asciinema.org
November 11, 2025 at 11:26 AM
Zet CLI aplikacija. Mislim da samo mene veseli, ali neka. :)
Reposted by Senko Rašić
OpenStreetMap Crash Course
Održat će se u Srijedu 05.11. u 19h na:
www.twitch.tv/turing_compl...
Sadržaj:
- uvod u OSM, kratka prezentacija s primjerima
- primjer rješavanja konkretnih zadataka
Trajanje cca 1h. Svi su dobrodošli naučiti malo o mapiranju.
bezdomni.net/osm-crash-co...
Održat će se u Srijedu 05.11. u 19h na:
www.twitch.tv/turing_compl...
Sadržaj:
- uvod u OSM, kratka prezentacija s primjerima
- primjer rješavanja konkretnih zadataka
Trajanje cca 1h. Svi su dobrodošli naučiti malo o mapiranju.
bezdomni.net/osm-crash-co...
OpenStreetMap Crash Course 2025
bezdomni.net
November 1, 2025 at 8:21 AM
OpenStreetMap Crash Course
Održat će se u Srijedu 05.11. u 19h na:
www.twitch.tv/turing_compl...
Sadržaj:
- uvod u OSM, kratka prezentacija s primjerima
- primjer rješavanja konkretnih zadataka
Trajanje cca 1h. Svi su dobrodošli naučiti malo o mapiranju.
bezdomni.net/osm-crash-co...
Održat će se u Srijedu 05.11. u 19h na:
www.twitch.tv/turing_compl...
Sadržaj:
- uvod u OSM, kratka prezentacija s primjerima
- primjer rješavanja konkretnih zadataka
Trajanje cca 1h. Svi su dobrodošli naučiti malo o mapiranju.
bezdomni.net/osm-crash-co...
Reposted by Senko Rašić
Evo ako je itko zainteresiran malo poslušati o otvorenim kartama, pripremit ću kratko predavanje i demo cca 1h, idući tjedan jednu večer. Napravio sam poll pa se upišite kad možete. Svi dobrodošli, ne morate nužno išta raditi kasnije. :)
strawpoll.com/3RnYXaxBmye
strawpoll.com/3RnYXaxBmye
October 28, 2025 at 8:33 AM
Evo ako je itko zainteresiran malo poslušati o otvorenim kartama, pripremit ću kratko predavanje i demo cca 1h, idući tjedan jednu večer. Napravio sam poll pa se upišite kad možete. Svi dobrodošli, ne morate nužno išta raditi kasnije. :)
strawpoll.com/3RnYXaxBmye
strawpoll.com/3RnYXaxBmye
"It's hard to get someone to fix a bug when their bottom line benefits from the bug not being fixed.", to paraphrase a popular quip.
Delta's version of "we get a lot of 500s, don't really care about fixing these ourselves; let's put the burden on the users"
October 25, 2025 at 7:20 PM
"It's hard to get someone to fix a bug when their bottom line benefits from the bug not being fixed.", to paraphrase a popular quip.
Reposted by Senko Rašić
![Comic. Two-column table. Column A lists planets: Mercury, Venus, Earth, Mars, Jupiter, Saturn, Uranus, Neptune. Column B has checkmark boxes for whether the planets have rings: the outer planets’ boxes are checked, and Earth has a red check. [Earth with a ring with an array pf points around it. An arrow labeled “Geostationary Satellite Bel”t points to the ring.] [caption] Astronomy fact: A century ago, Earth didn’t have rings, but we have one now! It’s where all the satellite dishes are pointed.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:cz73r7iyiqn26upot4jtjdhk/bafkreifauh76r3os5nnzjvkcz5sl2pgx5iz4sgkxebqjw3aogjn6e677aa@jpeg)
October 20, 2025 at 10:44 PM
Succintly put: "AI is writing 90% of the code [...] At the same time, AI doesn't own the code, [I] carry the responsibility for how it runs in production."
As an aside, happy to see I'm not the only maniac using AI to write raw SQL where I used to use ORM: github.com/senko/cijene...
As an aside, happy to see I'm not the only maniac using AI to write raw SQL where I used to use ORM: github.com/senko/cijene...
“Is 90% of code going to be written by AI? I don’t know. What I do know is, that for me, on this project, the answer is already yes.” lucumr.pocoo.org/2025/9/29/90...
90%
AI is writing 90% of the code I was in charge of
lucumr.pocoo.org
September 29, 2025 at 11:24 AM
Succintly put: "AI is writing 90% of the code [...] At the same time, AI doesn't own the code, [I] carry the responsibility for how it runs in production."
As an aside, happy to see I'm not the only maniac using AI to write raw SQL where I used to use ORM: github.com/senko/cijene...
As an aside, happy to see I'm not the only maniac using AI to write raw SQL where I used to use ORM: github.com/senko/cijene...
Reposted by Senko Rašić
The AlchemyConf version of my "Tell me a Story" talk is finally out 🎉
content.subvisual.com/talk-roundta...
content.subvisual.com/talk-roundta...
Saša Jurić || Tell Me a Story - Content Hub Blog
content.subvisual.com
September 29, 2025 at 8:53 AM
The AlchemyConf version of my "Tell me a Story" talk is finally out 🎉
content.subvisual.com/talk-roundta...
content.subvisual.com/talk-roundta...
Neka pati koga smeta, svaka bi ih majka za ZETa!
(za kontekst, natječe se 25 država)
(za kontekst, natječe se 25 država)
It's over! Nice to see a world championship where the USA and China didn't dominate the rankings
September 13, 2025 at 5:38 PM
Neka pati koga smeta, svaka bi ih majka za ZETa!
(za kontekst, natječe se 25 država)
(za kontekst, natječe se 25 država)
Next, people are going to call buying used books "stealing" because the authors get no money at all from the secondary sales.
I better sell a few of mine that I don't like enough to keep forever, before it gets outlawed as "piracy".
I better sell a few of mine that I don't like enough to keep forever, before it gets outlawed as "piracy".
The key detail people may miss: it looks like an AI company in the USA can train on an author's book by purchasing a used copy, cutting it up and scanning the pages - in which case the author gets no money at all!
September 6, 2025 at 7:30 PM
Next, people are going to call buying used books "stealing" because the authors get no money at all from the secondary sales.
I better sell a few of mine that I don't like enough to keep forever, before it gets outlawed as "piracy".
I better sell a few of mine that I don't like enough to keep forever, before it gets outlawed as "piracy".
Reposted by Senko Rašić
Reposted by Senko Rašić
Those would simply have never gotten written before. They don't get committed. They're useful but not earth-shattering. They contain a ton of code, especially if Claude also generates tests for them. And no one's job is getting replaced by them, nor do they devalue proper software development.
August 30, 2025 at 1:20 PM
Those would simply have never gotten written before. They don't get committed. They're useful but not earth-shattering. They contain a ton of code, especially if Claude also generates tests for them. And no one's job is getting replaced by them, nor do they devalue proper software development.
Reposted by Senko Rašić
When you use coding agents for something that produces the bulk of the code of an application that will be used for years, also factor in the technical debt that you are happily accumulating. When you use LLMs an an aid, you could, on the contrary, improve your coding culture.
August 21, 2025 at 1:08 PM
When you use coding agents for something that produces the bulk of the code of an application that will be used for years, also factor in the technical debt that you are happily accumulating. When you use LLMs an an aid, you could, on the contrary, improve your coding culture.
Clearly, they are prompting it wrong.
MIT’s NANDA initiative found that 95% of generative AI deployments fail after interviewing 150 execs, surveying 350 workers, and analyzing 300 projects. The real “productivity gains” seem to come from layoffs and squeezing more work from fewer people not AI.
MIT report: 95% of generative AI pilots at companies are failing
There’s a stark difference in success rates between companies that purchase AI tools from vendors and those that build them internally.
fortune.com
August 20, 2025 at 9:35 AM
Clearly, they are prompting it wrong.
There is a non-zero risk of AI apocalypse in which LLMs, rather than improving productivity (or at least maxxing paperclips), consume all the world's resources trying to produce anatomically and physically correct SVG of a pelican riding a bicycle.
And it will all be @simonwillison.net 's fault.
And it will all be @simonwillison.net 's fault.
Qwen3-4B-Thinking (a 4GB download) just became the first model I've tested to directly push back against my pelican riding a bicycle test, calling it "oddly specific and completely unrealistic" and saying "this request violates physics and biology"
simonwillison.net/2025/Aug/10/...
simonwillison.net/2025/Aug/10/...
August 11, 2025 at 8:45 PM
There is a non-zero risk of AI apocalypse in which LLMs, rather than improving productivity (or at least maxxing paperclips), consume all the world's resources trying to produce anatomically and physically correct SVG of a pelican riding a bicycle.
And it will all be @simonwillison.net 's fault.
And it will all be @simonwillison.net 's fault.
Reposted by Senko Rašić
Finally recorded something outside of the "studio" 😅 For this occasion I fittingly reached back to a 90s trip-hop classic by Morcheeba. Enjoy 🌊⛵🌞😎
youtu.be/FrdcxRRAibA
youtu.be/FrdcxRRAibA
The Sea (Morcheeba), fingerstyle
YouTube video by downcase string
youtu.be
August 11, 2025 at 9:23 AM
Finally recorded something outside of the "studio" 😅 For this occasion I fittingly reached back to a 90s trip-hop classic by Morcheeba. Enjoy 🌊⛵🌞😎
youtu.be/FrdcxRRAibA
youtu.be/FrdcxRRAibA
Reposted by Senko Rašić
Recorded some updated thoughts on agentic coding tools if someone is curious. www.youtube.com/watch?v=xfm9...
Agentic Coding Ecosystem 2025
YouTube video by Armin Ronacher
www.youtube.com
August 8, 2025 at 8:16 PM
Recorded some updated thoughts on agentic coding tools if someone is curious. www.youtube.com/watch?v=xfm9...
The issue of sky-high expectations from GPT-5 (as highlighted by @filippo.abyssdomain.expert) is that OpenAI have been terrible at managing expectations.
If you hype up "AGI achieved internally" and "PhD-level intelligence", users WILL expect miracles.
Compare with Anthropic or Chinese labs comms.
If you hype up "AGI achieved internally" and "PhD-level intelligence", users WILL expect miracles.
Compare with Anthropic or Chinese labs comms.
August 8, 2025 at 3:20 PM
The issue of sky-high expectations from GPT-5 (as highlighted by @filippo.abyssdomain.expert) is that OpenAI have been terrible at managing expectations.
If you hype up "AGI achieved internally" and "PhD-level intelligence", users WILL expect miracles.
Compare with Anthropic or Chinese labs comms.
If you hype up "AGI achieved internally" and "PhD-level intelligence", users WILL expect miracles.
Compare with Anthropic or Chinese labs comms.
Reposted by Senko Rašić
"I'm just saying," the first witch said. "A cat familiar is more traditional."
"Or a raven," the second witch said. "Where did she even get a dragon from, anyway?"
"She tamed it. In her last job," the third witch said.
"What job was that?"
"Princess."
"Oh. Well. No wonder she changed careers."
"Or a raven," the second witch said. "Where did she even get a dragon from, anyway?"
"She tamed it. In her last job," the third witch said.
"What job was that?"
"Princess."
"Oh. Well. No wonder she changed careers."
August 7, 2025 at 2:25 PM
"I'm just saying," the first witch said. "A cat familiar is more traditional."
"Or a raven," the second witch said. "Where did she even get a dragon from, anyway?"
"She tamed it. In her last job," the third witch said.
"What job was that?"
"Princess."
"Oh. Well. No wonder she changed careers."
"Or a raven," the second witch said. "Where did she even get a dragon from, anyway?"
"She tamed it. In her last job," the third witch said.
"What job was that?"
"Princess."
"Oh. Well. No wonder she changed careers."
Reposted by Senko Rašić
OpenAI released their long-promised open weight models today under clean Apache 2 licenses and with benchmarks that put them shockingly close to o3-mini and o4-mini
I've run the smaller (20B) model on my Mac and it's very impressive, despite only using ~15GB of RAM simonwillison.net/2025/Aug/5/g...
I've run the smaller (20B) model on my Mac and it's very impressive, despite only using ~15GB of RAM simonwillison.net/2025/Aug/5/g...
OpenAI’s new open weight (Apache 2) models are really good
The long promised OpenAI open weight models are here, and they are very impressive. They’re available under proper open source licenses—Apache 2.0—and come in two sizes, 120B and 20B. OpenAI’s …
simonwillison.net
August 5, 2025 at 8:39 PM
OpenAI released their long-promised open weight models today under clean Apache 2 licenses and with benchmarks that put them shockingly close to o3-mini and o4-mini
I've run the smaller (20B) model on my Mac and it's very impressive, despite only using ~15GB of RAM simonwillison.net/2025/Aug/5/g...
I've run the smaller (20B) model on my Mac and it's very impressive, despite only using ~15GB of RAM simonwillison.net/2025/Aug/5/g...
Reposted by Senko Rašić
Personally, I hope AI might kill (today's) copyrights and patents. It's probably one of my longest held beliefs that they are bad for innovation. At the same time such a path needs to be walked carefully and I appreciate what cloudflare is doing here.
x.com/eastdakota/s...
x.com/eastdakota/s...
Matthew Prince 🌥 on X: "Some supposedly “reputable” AI companies act more like North Korean hackers. Time to name, shame, and hard block them." / X
Some supposedly “reputable” AI companies act more like North Korean hackers. Time to name, shame, and hard block them.
x.com
August 4, 2025 at 11:21 PM
Personally, I hope AI might kill (today's) copyrights and patents. It's probably one of my longest held beliefs that they are bad for innovation. At the same time such a path needs to be walked carefully and I appreciate what cloudflare is doing here.
x.com/eastdakota/s...
x.com/eastdakota/s...
Reposted by Senko Rašić
OpenAI launched a "study mode" for ChatGPT today, and it appears to be almost entirely implemented as a system prompt
Thankfully OpenAI mostly don't take measures to protect their system prompt these days so it's easy to extract it and see how it works simonwillison.net/2025/Jul/29/...
Thankfully OpenAI mostly don't take measures to protect their system prompt these days so it's easy to extract it and see how it works simonwillison.net/2025/Jul/29/...
July 29, 2025 at 7:32 PM
OpenAI launched a "study mode" for ChatGPT today, and it appears to be almost entirely implemented as a system prompt
Thankfully OpenAI mostly don't take measures to protect their system prompt these days so it's easy to extract it and see how it works simonwillison.net/2025/Jul/29/...
Thankfully OpenAI mostly don't take measures to protect their system prompt these days so it's easy to extract it and see how it works simonwillison.net/2025/Jul/29/...
Reposted by Senko Rašić
The reality of offering “unlimited” plans for any service where the provider has costs based on usage.
There will always be users who abuse it in ways the provider doesn’t expect it (at least at first).
For AI, generating tokens costs $$. So “unlimited” is either subsidized or removed:
There will always be users who abuse it in ways the provider doesn’t expect it (at least at first).
For AI, generating tokens costs $$. So “unlimited” is either subsidized or removed:
July 29, 2025 at 6:58 AM
The reality of offering “unlimited” plans for any service where the provider has costs based on usage.
There will always be users who abuse it in ways the provider doesn’t expect it (at least at first).
For AI, generating tokens costs $$. So “unlimited” is either subsidized or removed:
There will always be users who abuse it in ways the provider doesn’t expect it (at least at first).
For AI, generating tokens costs $$. So “unlimited” is either subsidized or removed:
Reposted by Senko Rašić
I'm seeing a lot of commentary blaming the egregious data leak from the Tea dating safety app on vibe coding
I'm confident that, in this particular case, that's not what happened: the code at fault looks to have been written back in late 2023 simonwillison.net/2025/Jul/26/...
I'm confident that, in this particular case, that's not what happened: the code at fault looks to have been written back in late 2023 simonwillison.net/2025/Jul/26/...
Official statement from Tea on their data leak
Tea is a dating safety app for women that lets them share notes about potential dates. The other day it was subject to a truly egregious data leak caused by …
simonwillison.net
July 26, 2025 at 4:23 PM
I'm seeing a lot of commentary blaming the egregious data leak from the Tea dating safety app on vibe coding
I'm confident that, in this particular case, that's not what happened: the code at fault looks to have been written back in late 2023 simonwillison.net/2025/Jul/26/...
I'm confident that, in this particular case, that's not what happened: the code at fault looks to have been written back in late 2023 simonwillison.net/2025/Jul/26/...
This presumes "our" devices, "our" software, the services we subscribe to, work in for our benefit. In fact, they work for the benefit of shareholders.
Some call it enshitification, others call it late-stage capitalism.
Regardless of the name, we are mere users.
Some call it enshitification, others call it late-stage capitalism.
Regardless of the name, we are mere users.
I'll believe in AGI when my iPhone can filter out political spam text messages. But not before.
July 28, 2025 at 6:10 PM
This presumes "our" devices, "our" software, the services we subscribe to, work in for our benefit. In fact, they work for the benefit of shareholders.
Some call it enshitification, others call it late-stage capitalism.
Regardless of the name, we are mere users.
Some call it enshitification, others call it late-stage capitalism.
Regardless of the name, we are mere users.