rubslopes

> I can imagine a future in which some or even most software is developed by witches, who construct elaborate summoning environments, repeat special incantations (“ALWAYS run the tests!”), and invoke LLM daemons who write software on their behalf. These daemons may be fickle, sometimes destroying one’s computer or introducing security bugs, but the witches may develop an entire body of folk knowledge around prompting them effectively—the fabled “prompt engineering”. Skills files are spellbooks.

https://aphyr.com/posts/418-the-future-of-everything-is-lies...

But I considered that when I said "Good open weights models already exist and people can run them locally."

You can have a great LLM model with vast coding knowledge running on your computer right now, for free. It won't be the best one nor the fastest one, but still a very good one.

> Humans aren't benefiting from LLMs, only a few individuals are.

Honest question: how is this different from traditional Open Source? Linux powers most of the internet, yet the biggest beneficiaries are cloud providers, not individual users. Good open weights models already exist and people can run them locally. The gap between "open" and "everyone benefits equally" has always been there...

My setup is similar: I use hyperkey (caps lock) + letters to switch between apps, that are (almost) always maximized. So I'm always focusing on one app at a time.

I don't use this one, but a simpler one, also running on a vps. I communicate via telegram.

I say to it: check my pending tasks on Todoist and see if you can tackle on of those by yourself.

It then finds some bugs in a webapp that I took note. I tell it to go for it, but use a new branch and deploy it on a new url. So it clones the repo, fix it, commit, push, deploy, and test. It just messages me afterwards.

This is possible because it has access to my todoist and github and several other services.

I agree. No company is perfect, but if someone asked me to name the most consumer-friendly large tech company, I'd say Valve. And honestly, I can't think of a second one.

Carcinisation

I've tried the "4 agents running at the same time in different projects/features" and I felt literally dizzy. I still do the "check something else while the agent runs", and I often forget about that terminal window for many minutes, only to remember about it several tasks later.

The flip side of the coin is that if you don't have a determined rest time, you are always working. During my PhD, I couldn't feel the difference between weekdays and weekends; thus, I felt guilty on weekends when I was not making progress.

> The big players are just awful at marketing; too many SKUs and models - it takes a paragraph to figure out how 2 Dell laptops from the same release year differ.

Yes!! It's awful. I'm a long time Mac user and my wife needs a Windows laptop because of a specific software. I've tried three times to pick a computer for her, but I always give up after 10min and postpone the task...

Well, at least this one disclosed it...

But OpenClaw is "Claude Code" with bells and whistles so it can be contacted via messaging services and be woken up to do things at specific times.

Well, I, for one, am guilty of developing AI agents for public project evaluation for current governments. It's coming.

My experience also. I could manually connect my Obsidian notes to my AI, sure, but what I did instead was writing "Obsidian just released a CLI headless sync tool, install it so we can use it" and in a minute it came back with "Ok, everything installed, I just need your login and password."

Dangerous? Yes, very, but it truly feels like living in the future. Surprisingly, it's even more fun that sci-fi movies made me think this would be.

Ha! Just yesterday I set up a git repo to sync my Obsidian vault with my Ubuntu VPS for LLM use. Part of me wishes this had come out one day sooner, though honestly, I've grown to like the git workflow. The deal-breaker is mobile: it just doesn't play nicely there, so I'll keep using native sync for that.

Tools like OpenClaw have two core capabilities: the ability to rewrite themselves, and the ability to independently figure out how to connect to different services and establish those connections.

Yesterday, I was responding to a client ticket about what I knew wasn't a bug. It was something the client had requested themselves. The product is complex, constantly evolving, and has spawned dozens of related Jira tickets over time. So I asked my agent to explore the git history, identify changes to that specific feature, and cross-reference them with comments across the related tickets. Within minutes, I had everything I needed to write a clear response. It even downloaded PDF and DOCX files the client had attached. All of this was possible because my agent is connected to GitHub and Jira, and can clone repos locally since it runs on a VPS.

A second example: I was in an online meeting, taking notes as we went. Afterward, I asked the agent to pull the meeting transcript from Fireflies and use it to enrich my notes in Obsidian. I could have also asked it to push my action items straight into Todoist.

I was having too many bugs using it with my phone, I gave up and went back to Termux

Like OP, I use the VPS/SSH/Tailscale combo to run Claude Code from my phone, and I've never felt comfortable with the *Claw alternatives. This one finally made me take the leap. Simple, elegant, I'm having fun running it through Telegram.

I do believe I'm more productive, but my company is not charging much more for it. I'm working the same hours. Maybe that's the reason.

I just had a meeting yesterday when someone from the customer support team vibe-coded a solution in a few hours. The boss said, "Let's just give this as a gift; this product is not our focus and I want to show them how AI makes us work fast."

It's from the Orange Catholic Bible, I think.

"Thou shalt not make a machine in the likeness of a human mind."

I agree with your sentiment, and it touches on one of the reasons I left academia for IT. Scientific research is preoccupied with finding the truth, which is beautiful but very stressful. If you're a perfectionist, you're always questioning yourself: "Did I actually find something meaningful, or is it just noise? Did I gaslight myself into thinking I was just exploring the data when I was actually p-hacking the results?" This took a real toll on my mental health.

Although I love science, I'm much happier building programs. "Does the program do what the client expects with reasonable performance and safety? Yes? Ship it."

I don't understand this sentiment. It may hold true for other LLM use cases (image generation, creative writing, summarizing large texts), but when it comes to coding specifically, Google is *always* behind OpenAI and Anthropic, despite having virtually infinite processing power, money, and being the ones who started this race in the first place.

Until now, I've only ever used Gemini for coding tests. As long as I have access to GPT models or Sonnet/Opus, I never want to use Gemini. Hell, I even prefer Kimi 2.5 over it. I tried it again last week (Gemini Pro 3.0) and, right at the start of the conversation, it made the same mistake it's been making for years: it said "let me just run this command," and then did nothing.

My sentiment is actually the opposite of yours: how is Google *not* winning this race?

That's almost that famous quote by William Gibson: "The future is already here — it's just not very evenly distributed."

I predict that now that coding has become a commodity, smart young people drawn to technical problem-solving will start choosing other career paths over programming. I just don't know which ones, since AI seems to be commoditizing every form of engineering work.

The way benchmarks for Composer have been presented since v1 feels unusually cautious. To users, that reads as “the model isn’t very good”.

Thank you for this comment. I meditated almost daily for years, but somewhere along the way life got hectic and the habit slipped. Time to revisit it.

EDIT: I wanted to add that I think it's a great time to get back to it because this mental fatigue has been leading me to migrate to more analog tools, like pen and paper for journaling and ditching my smartwatch for analog ones.

I used to lose myself in focused work for hours. That's changed. Now I'm constantly pulled away, and I've noticed the pattern: I send a prompt, wait for the reply, and drift into browsing. Without SelfControl blocking me, I can't seem to resist. I am certainly more productive with LLMs, but I also feel much more tired (and guilty) after a day of work.

I've been using Coolify for deploying my webapps for about 3 years, it has a good web UI.

I used to say that Sonnet 4.5 was all I would ever need, but now I exclusively use Opus...

It sounds that I'm joking, but I'm not -- would it be so weird if those surgeons wore diapers?