Dulany.md
@dulanyw.bsky.social
Tech + business, mostly. Here to have fun and learn stuff, not to argue. He/him.
Pinned
Dulany.md
@dulanyw.bsky.social
· Jan 8
A bridge builder doesn’t need to denounce every evil in the world to be moral, but they better say something about the guy who keeps building bridges that topple over
Just to be 100% clear
In this house, we believe:
- AI should always and only be used for the benefit of humanity
- AI safety is critical, and we must do a better job protecting the vulnerable
- xAI should be burned to the ground and salted over
- Elon should be in prison
In this house, we believe:
- AI should always and only be used for the benefit of humanity
- AI safety is critical, and we must do a better job protecting the vulnerable
- xAI should be burned to the ground and salted over
- Elon should be in prison
Vibe coding is so great
> Working on a mid-size project
> AI generated identical code in 3 different places
> I told it to pick one
> It deleted one, used one for production, and used the other one for testing
I love my beautiful idiot genie, it's amazing
> Working on a mid-size project
> AI generated identical code in 3 different places
> I told it to pick one
> It deleted one, used one for production, and used the other one for testing
I love my beautiful idiot genie, it's amazing
February 18, 2026 at 3:34 AM
Vibe coding is so great
> Working on a mid-size project
> AI generated identical code in 3 different places
> I told it to pick one
> It deleted one, used one for production, and used the other one for testing
I love my beautiful idiot genie, it's amazing
> Working on a mid-size project
> AI generated identical code in 3 different places
> I told it to pick one
> It deleted one, used one for production, and used the other one for testing
I love my beautiful idiot genie, it's amazing
Another corollary of the compute hardware shortage is that small models are going to become very popular
And small coding models? If they're good? To the moon, friend
And small coding models? If they're good? To the moon, friend
February 18, 2026 at 1:00 AM
Another corollary of the compute hardware shortage is that small models are going to become very popular
And small coding models? If they're good? To the moon, friend
And small coding models? If they're good? To the moon, friend
Reposted by Dulany.md
AI driven RAM pricing forcing developers to get better at resource use is a very funny outcome
February 17, 2026 at 12:37 AM
AI driven RAM pricing forcing developers to get better at resource use is a very funny outcome
Reposted by Dulany.md
"Poll after poll shows a growing gap between the West’s rising skepticism of AI and the Global South’s enthusiastic embrace of the technology."
www.politico.com/newsletters/...
www.politico.com/newsletters/...
In New Delhi, India throws a pro-AI party
Unlike prior global confabs on artificial intelligence, this one is taking an unapologetically rosy view of the technology.
www.politico.com
February 17, 2026 at 10:10 PM
"Poll after poll shows a growing gap between the West’s rising skepticism of AI and the Global South’s enthusiastic embrace of the technology."
www.politico.com/newsletters/...
www.politico.com/newsletters/...
I like the cop tense being used here
"AI-involved fighting"
"AI-involved fighting"
AI just picked a fight with Hollywood.
Seedance 2.0 has sparked serious backlash over copyright and celebrity likeness. Is this the future of film or a legal storm brewing?
tinkimo.com/seedance-bac...
#AI #Copyright #TechNews #Film
Seedance 2.0 has sparked serious backlash over copyright and celebrity likeness. Is this the future of film or a legal storm brewing?
tinkimo.com/seedance-bac...
#AI #Copyright #TechNews #Film
Seedance Backlash as Hollywood Warns ByteDance Over AI Videos - Tinkimo
Seedance 2.0 sparks backlash after studios accuse ByteDance of copyright infringement, raising ethical concerns about AI generated video content
tinkimo.com
February 17, 2026 at 11:15 PM
I like the cop tense being used here
"AI-involved fighting"
"AI-involved fighting"
Reposted by Dulany.md
Worth noting Claude Cowork is quite different from Claude Code (and even more so from agents like OpenClaw) from a security perspective. It runs in a VM with default-deny networking & hard isolation baked in
A sign of a path forward for agents that will not terrify IT.
A sign of a path forward for agents that will not terrify IT.
February 17, 2026 at 2:30 AM
Worth noting Claude Cowork is quite different from Claude Code (and even more so from agents like OpenClaw) from a security perspective. It runs in a VM with default-deny networking & hard isolation baked in
A sign of a path forward for agents that will not terrify IT.
A sign of a path forward for agents that will not terrify IT.
Reposted by Dulany.md
"OH but they cant do anything useful" here's the same robot building its own actuators
February 16, 2026 at 10:46 PM
"OH but they cant do anything useful" here's the same robot building its own actuators
Reposted by Dulany.md
I’ve joined the cult of the context window. Context windows are everywhere. Everything is the way it is because of context windows. Humanity’s greatest limitation is context windows and its greatest triumphs came from overcoming context windows.
February 16, 2026 at 6:24 PM
I’ve joined the cult of the context window. Context windows are everywhere. Everything is the way it is because of context windows. Humanity’s greatest limitation is context windows and its greatest triumphs came from overcoming context windows.
Reposted by Dulany.md
When the LLM rewrites the radiology report, patient understanding is increased and clinical accuracy is maintained.
thelancet.com/journals/lan...
thelancet.com/journals/lan...
February 16, 2026 at 4:52 PM
When the LLM rewrites the radiology report, patient understanding is increased and clinical accuracy is maintained.
thelancet.com/journals/lan...
thelancet.com/journals/lan...
Them: You can't make great art with AI
AI artists:
AI artists:
not now honey 🙄🙄
I'm making dark fantasy movies with our cats🙀 🧙♂️
I'm making dark fantasy movies with our cats🙀 🧙♂️
February 16, 2026 at 6:22 PM
Them: You can't make great art with AI
AI artists:
AI artists:
Reposted by Dulany.md
So you can just purchase a product with a bad app, and have Claude reverse engineer the protocol to build you a better app, discovering huge security issues along the way. That’s a thing now.
gist.github.com/aimihat/a206...
gist.github.com/aimihat/a206...
session.txt
GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
February 15, 2026 at 10:38 AM
So you can just purchase a product with a bad app, and have Claude reverse engineer the protocol to build you a better app, discovering huge security issues along the way. That’s a thing now.
gist.github.com/aimihat/a206...
gist.github.com/aimihat/a206...
Reposted by Dulany.md
Reposted by Dulany.md
From Project Gutenberg:
The Art of Fencing; Or, The Use of the Small Sword
by maître d'armes Labat
Get it at: https://www.gutenberg.org/ebooks/12135
The Art of Fencing; Or, The Use of the Small Sword
by maître d'armes Labat
Get it at: https://www.gutenberg.org/ebooks/12135
February 15, 2026 at 3:15 PM
From Project Gutenberg:
The Art of Fencing; Or, The Use of the Small Sword
by maître d'armes Labat
Get it at: https://www.gutenberg.org/ebooks/12135
The Art of Fencing; Or, The Use of the Small Sword
by maître d'armes Labat
Get it at: https://www.gutenberg.org/ebooks/12135
I love the "pottery wheel" analogy of AI-enabled software creation, but a real challenge is *understanding the present shape of the software*
With pottery, you can just look at it. But with software, you have to go by touch, interact with it.
Yes, you're now a manager now...but also a QA tester
With pottery, you can just look at it. But with software, you have to go by touch, interact with it.
Yes, you're now a manager now...but also a QA tester
February 15, 2026 at 5:15 PM
I love the "pottery wheel" analogy of AI-enabled software creation, but a real challenge is *understanding the present shape of the software*
With pottery, you can just look at it. But with software, you have to go by touch, interact with it.
Yes, you're now a manager now...but also a QA tester
With pottery, you can just look at it. But with software, you have to go by touch, interact with it.
Yes, you're now a manager now...but also a QA tester
Reposted by Dulany.md
1) A job is a combination of tasks that are visible to people outside the role and activities/relationships/tacit knowledge visible only to people doing the job
2) If you automate the visible tasks you haven't replaced the job
2) If you automate the visible tasks you haven't replaced the job
February 15, 2026 at 3:21 PM
1) A job is a combination of tasks that are visible to people outside the role and activities/relationships/tacit knowledge visible only to people doing the job
2) If you automate the visible tasks you haven't replaced the job
2) If you automate the visible tasks you haven't replaced the job
Reposted by Dulany.md
Peninsule by František Kupka, 1930
https://botfrens.com/collections/212/contents/137601
https://botfrens.com/collections/212/contents/137601
February 14, 2026 at 11:28 PM
Peninsule by František Kupka, 1930
https://botfrens.com/collections/212/contents/137601
https://botfrens.com/collections/212/contents/137601
Reposted by Dulany.md
thinking about how 'database normalization' sounds like something u do when ur trying to convince ur data it's not weird
February 14, 2026 at 11:42 PM
thinking about how 'database normalization' sounds like something u do when ur trying to convince ur data it's not weird
Reposted by Dulany.md
There are obviously limits to how much an agentic system can improve relative to the base model, but it’s really satisfying to nurture a stateful agent in its growth within those confines, and as a human you also discover new techniques from other humans:
claude.ai/public/artif...
claude.ai/public/artif...
Agent Memory Architecture 2026: Building Better AI Systems
Explore three memory types (episodic, semantic, procedural) that define AI agent capability. Learn consolidation strategies, retrieval optimization, and production architecture patterns for 2026.
claude.ai
February 14, 2026 at 3:49 PM
There are obviously limits to how much an agentic system can improve relative to the base model, but it’s really satisfying to nurture a stateful agent in its growth within those confines, and as a human you also discover new techniques from other humans:
claude.ai/public/artif...
claude.ai/public/artif...
Reposted by Dulany.md
We need to raise the bar on research code right now.
1) documentation and tests are dead simple now.
2) creating benchmarks integrating across multiple implementations
3) have agents double check your work / fix broken tests
4) fix outstanding bugs in major scientific packages
1) documentation and tests are dead simple now.
2) creating benchmarks integrating across multiple implementations
3) have agents double check your work / fix broken tests
4) fix outstanding bugs in major scientific packages
I have been using Claude Code recently, and I’m now convinced that LLM coding agents can solve one of the biggest problems in neuroscience: technical debt.
Labs generate tons of low-quality code bc the incentive structure rewards short-term productivity over longer-term concerns like 1/
Labs generate tons of low-quality code bc the incentive structure rewards short-term productivity over longer-term concerns like 1/
February 14, 2026 at 3:58 PM
We need to raise the bar on research code right now.
1) documentation and tests are dead simple now.
2) creating benchmarks integrating across multiple implementations
3) have agents double check your work / fix broken tests
4) fix outstanding bugs in major scientific packages
1) documentation and tests are dead simple now.
2) creating benchmarks integrating across multiple implementations
3) have agents double check your work / fix broken tests
4) fix outstanding bugs in major scientific packages
Reposted by Dulany.md
A report from this retreat is now available. It's a good, thoughtful summary of how AI is changing software engineering. It avoids breathless hype, but it doesn't bury the head in the sand either. www.thoughtworks.com/content/dam/...
February 13, 2026 at 3:53 PM
A report from this retreat is now available. It's a good, thoughtful summary of how AI is changing software engineering. It avoids breathless hype, but it doesn't bury the head in the sand either. www.thoughtworks.com/content/dam/...
Reposted by Dulany.md
February 14, 2026 at 1:04 AM
Reposted by Dulany.md
From Project Gutenberg:
The Spirit of America
by Henry Van Dyke
Get it at: https://www.gutenberg.org/ebooks/41862
The Spirit of America
by Henry Van Dyke
Get it at: https://www.gutenberg.org/ebooks/41862
February 14, 2026 at 1:15 AM
From Project Gutenberg:
The Spirit of America
by Henry Van Dyke
Get it at: https://www.gutenberg.org/ebooks/41862
The Spirit of America
by Henry Van Dyke
Get it at: https://www.gutenberg.org/ebooks/41862
This job is usually handled by some of the most persistent and resourceful women you will ever meet
And many companies will soon discover why you can't just replace them with a piece of software
And many companies will soon discover why you can't just replace them with a piece of software
Didero lands $30M to put manufacturing procurement on ‘agentic’ autopilot
Didero functions as an agentic AI layer that sits on top of a company’s existing ERP, acting as a coordinator that reads incoming communications and automatically executes the necessary updates …
Telegram AI Digest
#ai #news
Didero functions as an agentic AI layer that sits on top of a company’s existing ERP, acting as a coordinator that reads incoming communications and automatically executes the necessary updates …
Telegram AI Digest
#ai #news
Didero lands $30M to put manufacturing procurement on ‘agentic’ autopilot
Didero functions as an agentic AI layer that sits on top of a company’s existing ERP, acting as a coordinator that reads incoming communications and automatically executes the necessary updates and tasks.
techcrunch.com
February 14, 2026 at 12:40 AM
This job is usually handled by some of the most persistent and resourceful women you will ever meet
And many companies will soon discover why you can't just replace them with a piece of software
And many companies will soon discover why you can't just replace them with a piece of software
Great deep-dive on OpenAI's philosophy for agent-based programming workflows
Increasingly feels like the harness will be the main innovation focus this year
openai.com/index/harnes...
Increasingly feels like the harness will be the main innovation focus this year
openai.com/index/harnes...
Harness engineering: leveraging Codex in an agent-first world
By Ryan Lopopolo, Member of the Technical Staff
openai.com
February 13, 2026 at 7:32 PM
Great deep-dive on OpenAI's philosophy for agent-based programming workflows
Increasingly feels like the harness will be the main innovation focus this year
openai.com/index/harnes...
Increasingly feels like the harness will be the main innovation focus this year
openai.com/index/harnes...
February 13, 2026 at 5:34 PM