Posts tagged: llm

All posts with the tag "llm"

94 posts latest post 2026-09-24
Publishing rhythm
Sep 2026 | 2 posts

I’m In On Agents

It’s the start of 2026 and agents are getting a lot better than they were. I’m using opencode at home, free mode with Zen and big pickle. At work I have access to a wider variety of models including what seems to be the gold standard 3 from anthropic opus, sonnet, haiku. Note I’ve sat on this post for over a month, refactored it a few times, things feel like they are moving so fast that its hard to keep my feelings straight. Things are changing # Around Aug 2025 I wrote I'm Out On Agents. I saw others in the space having such great success I gave it a solid shot, but found it to egregious edit more than I asked, make massive unneeded changes, and make more small bugs hidden in the details than was worth it. This was just after the tipping point where not just hype bros were making the switch, but people I work with, people I trust and have made really good Software in their career. People like the creator of Flask Armin Ronacher. People who are really good at writing code and do not ne…
Oxide and Friends 1/5/2026 – Predictions 2026!! - YouTube www.youtube.com
I thought this was an interesting take from Simon. I’ve been hearing him consistently say there will be more demand for software engineering in the future. More companies will have the ability and need to deal with software applications, but fewer of us will be hand editing any code. I thought this was an interesting interaction in the clip. Uh Simon, what do you got for us? I’ve just got the one. I think the act of the the the job of being paid money to type code into a computer Yeah. will go the same way as punching punch cards. Okay. I do I think in six years time I do not think it will anyone will be paid to 80:56 — just do the thing where you type the code. Just type the code. Okay. I think software engineering will still be an enormous career. I just think the software engineers won’t be spending multiple hours of their day in a text — editor typing out syntax. 81:09 It will look like punching cards. I think — so. Yeah. Yeah. Interesting. In uh in six years. Um and but software — engineering still very much exists. I believe so. I I hope so. [laughter] I — very much hope so because I think the um the challenge of being a software — 81:23 engineer is not remembering how if wha…

This is opencode’s init prompt.

Please analyze this codebase and create an AGENTS.md file containing:
1. Build/lint/test commands - especially for running a single test
2. Code style guidelines including imports, formatting, types, naming conventions, error handling, etc.
The file you create will be given to agentic coding agents (such as yourself) that operate in this repository. Make it about 150 lines long.
If there are Cursor rules (in .cursor/rules/ or .cursorrules) or Copilot rules (in .github/copilot-instructions.md), make sure to include them.
If there's already an AGENTS.md, improve it if it's located in <dir>
Photoshop for text — Steph Ango stephango.com
Photoshop for text In the near future, transforming text over an entire document will become as commonplace as filtering images. Steph Ango · stephango.com While the non deterministic nature of llms scare the heck out of me in the sense of just cutting it loose on my writing. letting it go through all of my files and just edit them. I do like the idea of mundane tools like “desaturate”, “Gaussian blur”, evolving out of it for text. I don’t yet see this with the tools we have now, but it will be interesting to see them evolve.
Vibe code is legacy code | Val Town Blog blog.val.town
Vibe code is legacy code Code that nobody understands is tech debt blog.val.town Brilliantly said. Vibe coding is legacy code. It’s code that we forget exists. Code that no one touches, you replace it. If you touch it you are more likely to break it. The worst possible situation is to have a non-programmer vibe code a large project that they intend to maintain. This would be the equivalent of giving a credit card to a child without first explaining the concept of debt. As you can imagine, the first phase is ecstatic. I can wave this little piece of plastic in stores and take whatever I want! … Read more in the full post

When To Vibe Code

I enjoyed this post from Theo and think it deserves re-iterated, revisited, and to remind myself of some of these things. https://youtu.be/6TMPWvPG5GA?si=guQem4R8dLOMBntP&t=1356 The skill/read spectrum # The first diagram describes that there has become a spectrum of agentic coding from vibe coding where you don’t ready anything, to looking at everything in detail, across a group of people who don’t have a clue what the code says to people who could do it way better if they took the time. The importance spectrum # He argues here that its ok to bounce between A,B, and D, but C becomes dangerous. I’d argue that he brought up a 3rd spectrum that is important later on, “how critical is this”. I think the I don’t know, don’t care, didn’t read, but the thing did its job is quite fine, but don’t know, don’t care, mission critical is the main issue we are seeing with agentic coding, primarily in the didn’t read but critical Zone. The Rules # This is the list that prompted this post, I think it…
2 min read
aillm

I’m Out On Agents

Its the year 2025 and we are only a few years into having 6 months to live before ai takes our jobs, and the big push right now is agents, managing agents. I will fully concede to I’m not doing it right, or a future state gets better than where we are right now, but right now they kinda suck. Transparency I’m sitting offline right now as I write this, These are my feels, no research, no links, no ai, just vibes. Chat # Chat is what really kicked off ai uses and goes back as old as computers, but it always sucked. Then chatgpt rocked the world with the biggest launch day in history and showed us that it could actually be pretty good. Unethically trained on everything they could get their hands on, burning cities worth of electricity to train, and keep training to stay ahead of the competition. It does a damn good job. There are tells, and if you see enough of it there is a lot that turns to slop, but if you had never seen it before, there is no way you would assume that it was not a com…
The Death of the User Interface gist.github.com
THE_DEATH_OF_THE_USER_INTERFACE.md markdown This is an insane level of agentic llm use, the author claims to not even use his filesystem anymore, its too cumbersome to find where downloads and documents are and way too easy to ask an agent to move all pdf’s from downloads to documents. This scares me on multiple levels, theres the epidemic of ai datacenters and ai companies burning cash, burning through gpu’s and api calls to a giant data center just to move files sounds absolutely insane to me. Then there is the level of accuracy. There’s a level of interpretation that happens with english that does not happen in code, code generally does what it does repeatably. The examples of “Move old backups to archive” is so open for interpretation that its ready to ruin your day, which backup, which archive, how old? Dude is ready to loose his files. Last is privacy, there is going to be a privacy epidemic that is going to pwn so many people giving these things full access to email, chat, ALL of YOUR FILES, and computer, your whole network. Look I get it this is where things are going, but dude is early, like too early. These things need to become far more accurate, less power hungry, and r…
The Internet is Dying: AI, Bots, and The End of Human Content … www.youtube.com
Damn this VAnessa is hitting my feed with hard topics, I’m not sure whether to subscribe or to block. These top websites only feel worse every day, when I post on twitter and I get 4 likes by accounts that were created 5 minutes ago with racy profile pics it feels obvious. I wonder how larger accounts deal with it. Now that llms have made making these bots mimic humans easy It really makes you want out. I’ve really become a curmudgeon and leaning on rss over the past year, I dont like it, but idk what to do.
The Real Reason Your Power Bill Doubled (It’s Not Inflation) -… www.youtube.com
This is a crazy theory I did not realize was out there, but damn power just keeps costing more and more. She does not mention it here, but there are many sources of power for the grid that cost vastly different amounts to produce, generally “clean energy” solutions are harder and more expensive to bring online and don’t just turn on and off at the flick of a switch. Anyways, how are the power companies divying this power out to users, do some get preferred rates or supply? My rates just went up for the summer period “temporarily”. Our infrastructure is aging hard to upgrade and needs something done to it. Who’s really going to pay for it, these AI companies are throwing big numbers but do they have any real money? Do they have any real revenue after building out massive data centers filling them with the most expensive hardware? These guys are burning cash like crazy.
Slops | justin․searls․co justin.searls.co
Slops AI-generated slop that I thought was worth sharing. justin․searls․co · justin.searls.co Justin has such great feeds on his site, I love how the main feeds are so prominant just to the left of the article you are reading. in particular feels like a great category. Saving this chat for later, or found it particularly interesting, but don’t really want to make a post about it.
ThePrimeagen on X: “there is literally no universe that this i… x.com
ThePrimeagen (@ThePrimeagen) on X there is literally no universe that this is true 10k lines and its not bug filled crap? ok Lex Luthor, its time to step away from the keys X (formerly Twitter) · x.com there is literally no universe that this is true 10k lines and its not bug filled crap? ok Lex Luthor, its time to step away from the keys Is this 10k real production code? Dry in the sense that it hasn’t re-implemented the same s3 api dozens of time? What language are we talking something dense like python? something very verbose like html? Maybe a language where you implement everything from scratch like lua. This matters a lot. Playing with little POC applications that dont mean anything I can quickly come up with 500-1k likes of code that I may never look at again. I’m sure I can come up wtih 10k decent lines of code a day. But for the same application without duplicating everything over and over? For something that moves the needle and really matters?? every single day?? Consistently +10k, not 10k changes, not 10k deletes of yesterdays code. nah thats wack.
I Will Fucking Piledrive You If You Mention AI Again — Ludicity ludic.mataroa.blog
Fantastic write up on their experience in ai, opinions on ai being a hoax with a veil of reasonable usefulness. Arguing that most people do not understand enough to see the difference, and thought leaders see where it is now, see where it was yesterday, it must be going to general intelligence tomorrow and you all will loose your jobs without this. I appreciate the satirical language here.
Meet Gor | I kind of hate Agentic IDEs for the sake of product… www.meetgor.com
External Link meetgor.com Yes, I can review the code and make changes, but who in the world loves reviewing code? Do you love reviewing peers’ PRs? really? I’m with MeetGor here 100%. reviewing the nuance, not being as involved with the process of creating the architecture design, not solving the problems that arise in development make it hard to effectively review and not turn into LGTM man.

I’ve been using gitingest web ui [[ thoughts-516 ]] for quite awhile to serialize git repo into llm friendly text files. This gives tools context about repos that are not in the training data so that it knows about it and how to use the code in the repo. gitingest also has a python library [[ thoughts-517 ]]

I had a use case for a project not yet on git, and found yek.

Installing yek #

Their instructions tell you to curl to bash.

curl -fsSL https://bodo.run/yek.sh | bash

I don’t like curl to bash from random sites, so I have my own self hosted version of i.jpillora.com. I like using this because it pulls from github and I trust github as a source for artifacts as good as the repo I am pulling from.

curl https://i.jpillora.com/bodo-run/yek | bash

Using yek #

yek

/tmp/yek-output/yek-output-bb01e621.txt

This will give you a link to a text file that you can add to many llm tools. This happened so fast for me that I didn’t even believe that it worked properly.

more options #

As with most clis, you can run yek --help to see the options available.

yek --help
What ChatGPT is NOT - Tech Raven Blog blog.techravenconsulting.com
What ChatGPT is NOT - Tech Raven Blog There is a lot of excitement about ChatGPT and how it allows us to interact with information and technology. I am actually excited that it now exists and still, I think it is being way overhyped. I... Tech Raven Blog - · blog.techravenconsulting.com Do you remember regression models from college: given some data, you find a best fit line that allows you to predict Y given X. At the end of the day, ChatGPT, and LLMs in general, are the same thing as the regression model – it’s just that ChatGPT is the largest and fanciest model we currently have to model language and information. I really am coming to the idea of calling it a “word calculator”, this seems to be the most succinct description of llms that the lay person can comprehend and relate to. ChatGPT does not hallucinate or become unhinged I think Steve goes much deeper on this in his intervew on fafo.fm. They describe it more as a pleaser or “yes man” essentially all the companies that are building these models want to give the “best” answer, better than their competitors. With this comes the risk of it being completely wrong, they are designed to always give an answer. On that episode the…
The Rule - YouTube www.youtube.com
Under 2000 everything is happy, green field. Any decision you have made is relatively easy to back out of (barring you making a library with downstream users), but as you go, regret kicks in. Regret we didn’t make that pydantic 2 upgrade earlier, as new features become more apealing. Regret that we chose sqlite for simplicity, speed, agility, and now we might need robust and distributed. Regret that you chose a front end framework, or to have a front end at all to a backend problem. Regret that you put 6 layers of abstraction on your db early on and now that you understand the problem you want different abstractions, but all of your endpoints deeply depend on the current one. Vibe coding will not save you, it will only make these wrong decisions for you without the context that you have. You will hate it’s decisions more because you had no input into some of them.
Vectorizing Your Databases with Steve Pousty www.fafo.fm
Vectorizing Your Databases with Steve Pousty What exactly is an LLM doing and why do you need to learn so many new terms? Steve Pousty is here to explain that most of those new terms are things you already kno… Fork Around And Find Out · fafo.fm Steve is such a great listen, the neurospicy 🌶️ rambles this episode goes on is so relatable. I feel like I really missed out on some great takes on intellij vs neovim, but got some really great knowledge about vector db’s, embedding, text compression, similarities to vector algegra like infinite craft. Just popped open infinitecraft and I’ve definitely played this with my kids before, super fun, just could not remember the name of this one. I do remember an android one as well that is alchemist or something like that, which we have also played a lot.
A quote from Mark Zuckerberg simonwillison.net
A quote from Mark Zuckerberg You also mentioned the whole Chatbot Arena thing, which I think is interesting and points to the challenge around how you do benchmarking. How do you know what models are … Simon Willison’s Weblog · simonwillison.net Interesting how confidently he says we can easily go to the top. really makes you wonder what we the normies are leaving on the table by using these general purpose models and what could be achieved with really tuned in models. Could I make an automatic blog tagger more accurately, maybe smaller, maybe tuned so well it runs fine on cpu?