I keep seeing posts like this, all the time https://www.reddit.com/r/ClaudeAI/comments/1q5lt9g/developer_uses_claude_code_and_has_an_existential/

I use Claude Code, daily. Yes it’s great. But it also consistently (although not very often) makes horrible decisions and writes dumbest code possible, which means it’s absolutely incapable of working on its own without meticoulus guidance, unless you want your project to be unusable mess. I love this tool because it speeds up development a lot but there are rarely days without making a facepalm when I see its mistakes.

8 yoe

I’m 45 years old, and I’ve been a software engineer for 22 years. Although I like doing software development, I’ve been laid off multiple times in my career, and I’m getting a bit tired of that. I’ve wondered what other jobs & careers I could easily get into at this point in my life and still have a decent salary? I realize I could take a salary cut if I do that, but I’m curious if I could easily get into a job that’s more stable that I’d still enjoy.

Also, it saddens me to feel this way. I feel like we need software developers & other tech workers, but I also feel like I wouldn’t recommend others go into this field anymore due to the lack of job stability.

I am hearing a lot of stuff about agents as in Claude Code or Antigravity. There are even people saying, that they have 5, 10 or even 20+ agents running at the same time while saying, that they haven’t touched an IDE for weeks. That would be a huge change in my workflow. So I gave it an honest try with Antigravity since I have the Gemini Subscriptions anyways and its included in the pricing.

I tried to bootstrap a project and generated specific prompts for the agent and it started working. It looked really good first and I was slowly getting further with the project. I could not “one-shot” every functionality, but I kept going further and further.

At some point, some features were not working as expected. I tried many ways around it, but then something else started to break.

So eventually I have hit a point, where I suddenly have some random codebase with many thousands of lines and nobody on this world knows how anything is working at all. And even if it seemingly …

Another “I’m feeling so fatigued by AI” post here.

I work at a finance company that pretends we’re a tech company. The last two years have led to what is now an insufferable culture. Every single OKR involves AI. We’re being spammed every day across Teams Channels with posts from senior leaders with AI propaganda. It’s being asked to be implemented into products where it makes no sense at all. All of our learning workshops and our budget has gone towards AI. They’ve openly stated that our biggest problem is the bottleneck of code reviews and are looking to automate that.

I’ve grown so bitter towards it mainly because it feels like we’ve entered the age of anti-intellectualism. We’re now apparently ecstatic that there’s no critical thinking required anymore. Hard-earned skills and knowledge are being democratized to people who quite frankly haven’t earned them.

This hurts because I’m a very driven person. I taught myself to become a software engineer without a degree. It took three …

Wondering if anyone knows of any companies that are going against the grain and are actively against AI use in their engineering and/or products. Any that are taking a big fat audacious bet against the AI trend? Seems like it would be a huge gamble but could also have a potentially huge upside if everyone else in the market going all in on AI for and in everything ends up crashing and burning. Genuinely curious if there are any examples of tech companies actively pursuing an anti-AI strategy.

This is an AI-assisted application where the system design and UX are implemented manually, with AI used as a runtime component.

I wanted an AI assistant that doesn’t live in the cloud or inside a browser.

So I built a small self-hosted system that runs locally and exists as a mirror in my room. You talk to it by voice, it responds by voice, and then it fades back into the background.

The idea was to give a local LLM a physical presence, not another UI.

It’s running on my own hardware (Raspberry Pi + local LLM stack), and the whole thing is open source.

It’s still early and rough in places, but the core interaction works.

I’m curious if anyone else here is interested in physical interfaces or non-screen-based ways of interacting with local AI.

GitHub: https://github.com/orangekame3/mirrormate

Features:

• Pi 4 (2GB RAM), in a Geekworm NASPI-lite case. Modified to fit a larger 2.5” 5tb HDD, 20000mah battery and added power/status led button.
• 5tb HDD, storing a mirrored/synced copy of my complete media library
• Two wifi adapters:
  • A) Connecting as guest to wifi for local LAN/internet access
  • B) Providing hotspot for administration and streaming to local devices (ie offline playback)
• HDMI output, for connecting directly to TVs and playing via Kodi (with Jellyfin plug-in). Repurposed Firestick remote control.
• Tailscale so it automatically syncs from the remote master library whenever it’s online

Weight: 2lbs. Running time: 10 hours, streaming 4k video Cost: $170

-——
Fyi: This replaces WD My Passport Wireless Pro 2TB, which had most of the same features.

The Passport:

• only 1.4 lbs
• 2tb drive
• Running a limited Debian Linux repo (last firmware update 2019
• No fileshare access controls, anyone on the wifi/LAN has write access
• No HDMI/local playback
• Plex …