Chad
@chadfennell.com
Code (Elixir & TS mostly), UX, Product Development
Pinned
Chad
@chadfennell.com
· 12d
When one socket closes another one opens.
At least if you are running on the BEAM.
At least if you are running on the BEAM.
Martin Kleppmann (Designing Data-Intensive Applications) on LLMs & their potential for unlocking formal verification (at last) martin.kleppmann.com/2025/12/08/a... I wrote some Rocq ages ago, neat but *non-trivial* barrier to entry. These systems prove the existence of subtle unhandled assumptions
Prediction: AI will make formal verification go mainstream — Martin Kleppmann’s blog
martin.kleppmann.com
February 11, 2026 at 12:12 AM
Martin Kleppmann (Designing Data-Intensive Applications) on LLMs & their potential for unlocking formal verification (at last) martin.kleppmann.com/2025/12/08/a... I wrote some Rocq ages ago, neat but *non-trivial* barrier to entry. These systems prove the existence of subtle unhandled assumptions
Cool, thanks for the heads-up!
February 10, 2026 at 10:41 PM
Cool, thanks for the heads-up!
That all sounds pretty familiar to me - similar exp. Not to plead "skill issue" here but there is also an art to getting a good result.
And interactive learning, refactoring suggestions/code reviews, etc are, to me, net wins for groking new things & QAing old ones. Discipline is still required!
And interactive learning, refactoring suggestions/code reviews, etc are, to me, net wins for groking new things & QAing old ones. Discipline is still required!
February 10, 2026 at 4:06 AM
That all sounds pretty familiar to me - similar exp. Not to plead "skill issue" here but there is also an art to getting a good result.
And interactive learning, refactoring suggestions/code reviews, etc are, to me, net wins for groking new things & QAing old ones. Discipline is still required!
And interactive learning, refactoring suggestions/code reviews, etc are, to me, net wins for groking new things & QAing old ones. Discipline is still required!
Charm is just delightful charm.land I want to build something with this pretty bad rn
Charm
We make the command line glamorous
charm.land
February 10, 2026 at 12:08 AM
Charm is just delightful charm.land I want to build something with this pretty bad rn
Checks out. Some related research - "employees worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day, often without being asked to do so" "...leading to workload creep, cognitive fatigue, burnout, and weakened decision-making"
hbr.org/2026/02/ai-d...
hbr.org/2026/02/ai-d...
AI Doesn’t Reduce Work—It Intensifies It
One of the promises of AI is that it can reduce workloads so employees can focus more on higher-value and more engaging tasks. But according to new research, AI tools don’t reduce work, they consisten...
hbr.org
February 9, 2026 at 9:00 PM
Checks out. Some related research - "employees worked at a faster pace, took on a broader scope of tasks, and extended work into more hours of the day, often without being asked to do so" "...leading to workload creep, cognitive fatigue, burnout, and weakened decision-making"
hbr.org/2026/02/ai-d...
hbr.org/2026/02/ai-d...
In Crocodile Dundee voice "That's not a cargo bike. This is a cargo bike." fastport.honda.com
Fastport Honda
fastport.honda.com
February 9, 2026 at 3:14 PM
In Crocodile Dundee voice "That's not a cargo bike. This is a cargo bike." fastport.honda.com
I've been kicking the tires on Incus (formerly LXD) to level up my agent dev env. You can have something like Fly.io's sprites on your own box - snapshots, isolation etc. Agent Incus is a few bash scripts I put together for better ergonomics github.com/MixTapeSoftw...
GitHub - MixTapeSoftware/agent_incus: Secure AI agent code execution sandbox using Incus containers with resource limits, network isolation, and unprivileged isolation for IA Agents and Development
Secure AI agent code execution sandbox using Incus containers with resource limits, network isolation, and unprivileged isolation for IA Agents and Development - MixTapeSoftware/agent_incus
github.com
February 8, 2026 at 3:56 PM
I've been kicking the tires on Incus (formerly LXD) to level up my agent dev env. You can have something like Fly.io's sprites on your own box - snapshots, isolation etc. Agent Incus is a few bash scripts I put together for better ergonomics github.com/MixTapeSoftw...
Reposted by Chad
I love having my biases validated dashbit.co/blog/why-eli...
Why Elixir is the best language for AI - Dashbit Blog
A recent study by Tencent showed that Elixir had the highest completion rate across models when compared among 20 different programming languages. In this article, we explore the reasons why that may ...
dashbit.co
February 5, 2026 at 4:45 PM
I love having my biases validated dashbit.co/blog/why-eli...
The Minnesota I knew and loved is under invasion. The response: regular people standing up for their neighbors, courageous moms with phones and whistles, and despite the murder of innocent people and knowing they could be next, love and salvation.
February 4, 2026 at 7:26 PM
The Minnesota I knew and loved is under invasion. The response: regular people standing up for their neighbors, courageous moms with phones and whistles, and despite the murder of innocent people and knowing they could be next, love and salvation.
Reposted by Chad
Another commission for a client's dark fantasy video game. The setting here is a dark spooky catacombs that leads into a scary encounter.
January 29, 2026 at 5:02 AM
Another commission for a client's dark fantasy video game. The setting here is a dark spooky catacombs that leads into a scary encounter.
When one socket closes another one opens.
At least if you are running on the BEAM.
At least if you are running on the BEAM.
January 29, 2026 at 7:24 PM
When one socket closes another one opens.
At least if you are running on the BEAM.
At least if you are running on the BEAM.
An experimental mitmproxy wrapper app to help lower the risk of exfiltration by tackling some low hanging fruit: block requests to known bad urls, request heuristics etc.
github.com/MixTapeSoftw...
github.com/MixTapeSoftw...
GitHub - MixTapeSoftware/agent_proxy
Contribute to MixTapeSoftware/agent_proxy development by creating an account on GitHub.
github.com
January 26, 2026 at 6:42 PM
An experimental mitmproxy wrapper app to help lower the risk of exfiltration by tackling some low hanging fruit: block requests to known bad urls, request heuristics etc.
github.com/MixTapeSoftw...
github.com/MixTapeSoftw...
Reposted by Chad
I built a light that can see radio waves
YouTube video by Rootkid
www.youtube.com
January 23, 2026 at 11:26 AM
Some truly cutting engineering ideas here - e.g. a raft consensus model for self-healing in the event of a disk failure for a given instance of a DB
www.youtube.com/watch?v=sC1B...
www.youtube.com/watch?v=sC1B...
The FASTEST and SAFEST Database
YouTube video by The PrimeTime
www.youtube.com
January 21, 2026 at 7:57 PM
Some truly cutting engineering ideas here - e.g. a raft consensus model for self-healing in the event of a disk failure for a given instance of a DB
www.youtube.com/watch?v=sC1B...
www.youtube.com/watch?v=sC1B...
Reposted by Chad
This is great - it's about time someone updated the discourse on LLM energy usage to reflect that coding agents use massively more prompts than occasional questions to ChatGPT
Simon estimates that a day of coding agent usage comes out close to the energy needed to run a dishwasher
Simon estimates that a day of coding agent usage comes out close to the energy needed to run a dishwasher
Whenever I read discourse on AI energy/water use that focuses on the "median query," I can't help but feel misled. Coding agents like Claude Code send hundreds of longer-than-median queries every session, and I run dozens of sessions a day.
On my blog: www.simonpcouch.com/blog/2026-01...
On my blog: www.simonpcouch.com/blog/2026-01...
January 20, 2026 at 11:10 PM
This is great - it's about time someone updated the discourse on LLM energy usage to reflect that coding agents use massively more prompts than occasional questions to ChatGPT
Simon estimates that a day of coding agent usage comes out close to the energy needed to run a dishwasher
Simon estimates that a day of coding agent usage comes out close to the energy needed to run a dishwasher
Reposted by Chad
Hey Elixir friends! :) Hologram's journey just took a big leap forward: @curiosum.bsky.social is coming on board as the Main Sponsor, and I'm joining their team to work on Hologram full-time!
Read what this means for Hologram: hologram.page/blog/hologra...
#Hologram #Elixir #ElixirLang #BEAM
Read what this means for Hologram: hologram.page/blog/hologra...
#Hologram #Elixir #ElixirLang #BEAM
January 21, 2026 at 1:39 AM
Hey Elixir friends! :) Hologram's journey just took a big leap forward: @curiosum.bsky.social is coming on board as the Main Sponsor, and I'm joining their team to work on Hologram full-time!
Read what this means for Hologram: hologram.page/blog/hologra...
#Hologram #Elixir #ElixirLang #BEAM
Read what this means for Hologram: hologram.page/blog/hologra...
#Hologram #Elixir #ElixirLang #BEAM
Fascinating gig!
I'm hiring!
For a pretty fun job, if you ask me.
Talk to some of the most interesting teams and companies across the globe, understand out what they do, then explain this to fellow software engineers. Full-remote, flexible €150+/$175+ per hour
Details: blog.pragmaticengineer.com/tech-industr...
For a pretty fun job, if you ask me.
Talk to some of the most interesting teams and companies across the globe, understand out what they do, then explain this to fellow software engineers. Full-remote, flexible €150+/$175+ per hour
Details: blog.pragmaticengineer.com/tech-industr...
January 20, 2026 at 5:42 PM
Fascinating gig!
"These findings differ from earlier studies that associated rudeness with poorer outcomes, suggesting that newer LLMs may respond differently to tonal variation."
January 20, 2026 at 3:37 PM
"These findings differ from earlier studies that associated rudeness with poorer outcomes, suggesting that newer LLMs may respond differently to tonal variation."
Great news for humanity - rude prompts do better than polite ones (for now):
"Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts"
arxiv.org/pdf/2510.04950
"Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts"
arxiv.org/pdf/2510.04950
arxiv.org
January 20, 2026 at 3:37 PM
Great news for humanity - rude prompts do better than polite ones (for now):
"Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts"
arxiv.org/pdf/2510.04950
"Contrary to expectations, impolite prompts consistently outperformed polite ones, with accuracy ranging from 80.8% for Very Polite prompts to 84.8% for Very Rude prompts"
arxiv.org/pdf/2510.04950
I am ecstatic to be on localhost for a day as an attendee with this cast of software legends! Join me!
See the full agenda and apply here, while seats are left: pragmaticsummit.com
I'm beyond impressed how this event is turning out to be - what a special crowd this will be (not just speakers, but attendees as well!)
I'm beyond impressed how this event is turning out to be - what a special crowd this will be (not just speakers, but attendees as well!)
The Pragmatic Summit
A one-day summit hosted by Gergely Orosz and The Pragmatic Engineer.
pragmaticsummit.com
January 14, 2026 at 11:32 PM
I am ecstatic to be on localhost for a day as an attendee with this cast of software legends! Join me!
🎵 vectors in high dimensional space
all we are is vectors in high dimensional space 🎵
Solid framing of current debates on language models in relation to language acquisition, language "understanding" etc
arxiv.org/abs/2401.03910
all we are is vectors in high dimensional space 🎵
Solid framing of current debates on language models in relation to language acquisition, language "understanding" etc
arxiv.org/abs/2401.03910
A Philosophical Introduction to Language Models -- Part I: Continuity With Classic Debates
Large language models like GPT-4 have achieved remarkable proficiency in a broad spectrum of language-based tasks, some of which are traditionally associated with hallmarks of human intelligence. This...
arxiv.org
January 14, 2026 at 10:31 PM
🎵 vectors in high dimensional space
all we are is vectors in high dimensional space 🎵
Solid framing of current debates on language models in relation to language acquisition, language "understanding" etc
arxiv.org/abs/2401.03910
all we are is vectors in high dimensional space 🎵
Solid framing of current debates on language models in relation to language acquisition, language "understanding" etc
arxiv.org/abs/2401.03910
Reposted by Chad
Whatever thoughts about the river have arisen now, they never describe the river at this moment.
The actual state of the river is indescribable.
It is eternally unique, just as it is, now.
The actual state of the river is indescribable.
It is eternally unique, just as it is, now.
January 6, 2026 at 8:09 PM
Whatever thoughts about the river have arisen now, they never describe the river at this moment.
The actual state of the river is indescribable.
It is eternally unique, just as it is, now.
The actual state of the river is indescribable.
It is eternally unique, just as it is, now.