cd /news/artificial-intelligence/recovering-from-agentic-coding · home topics artificial-intelligence article
[ARTICLE · art-82606] src=jrzs.dev ↗ pub= topic=artificial-intelligence verified=true sentiment=· neutral

Recovering from Agentic Coding

A software engineer who used AI extensively for coding reports that over-delegating to AI degraded his ability to write and reason about code, prompting him to build a Pong game from scratch using the Odin language and Raylib library without AI assistance. The experience rekindled his passion for low-level programming and computer science fundamentals, leading him to study database design, CPU caches, and memory alignment, and he is now working on a Tetris game.

read2 min views2 publishedJul 31, 2026
Recovering from Agentic Coding
Image: source

I’m very bullish on AI, but I’m not ignorant of its drawbacks.

Long term AI use is less about quickly delivering value to customers, but more about managing complexity.

The skill atrophy is real though. Over-delegating technical planning and implementation to AI really degraded my ability to write and reason about code.

I use AI to plan, write, and help review code. Each time I do, I feel weaker in those three areas.

A few weeks ago, I decided to challenge myself. Make a game by hand. Like a caveman, and avoid AI at all costs.

I picked a language I never heard of before, Odin, and a lightweight game library I’ve never used before, Raylib. This was a very humbling experience to say the least. I’ve never made a game before.

Considering I was used to whipping up UIs in seconds, going back to crawling at a snails pace was incredibly tough. I was poring over READMEs, hovering over every function and reading the LSP provided docs in detail, taking it all in and trying to bang rocks together to make fire.

As for the game itself, I kept it “simple” and decided to write Pong from scratch, which was much harder than I thought it would be. Especially without a game engine to structure the whole thing. I never really had to think about game loops, collisions, physics, vector math, or frame deltas before. As a backend engineer, this was all new to me. Much less how to program and organize a game project.

It didn’t take too long before things started to click and my programmer brain woke up. I was like my old self again, flying around my editor with vim keybindings, thinking deeply about code boundaries, and refactoring everything. It was exciting, empowering, and fun. Felt like if the internet were to go away tomorrow, I’d be alright.

This whole experience has also been a gateway drug to low level thinking for me. Thank you, Ginger Bill (creator of Odin). I’m finding myself now gravitating towards reading about database design, CPU caches and memory alignment. Systems Performance by Brendan Gregg has been a godsend for this rekindled love of computer science fundamentals.

I’ve no doubt there are others who feel the same way I did about AI, and question their self-worth and individual contribution. I think despite the atrophy, the skills are still there, it just takes a beat to wake them up and sharpen them again.

I’ve since moved onto making Tetris, which will take me some time before a playable version is ready end to end but I’m enjoying every minute of the process.

── more in #artificial-intelligence 4 stories · sorted by recency
── more on @odin 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/recovering-from-agen…] indexed:0 read:2min 2026-07-31 ·