Reply to Austin's comment How much do you look at the code if at all? If I'm vibe coding a tool or personal project I really don't want to read the code rather just reprompt

@starkstechnology

7.7K followers · usually 1.5K views a post

Views

5.6K

Likes

250

Comments

23

Shares

7

Selling · SaaS › AI Tool

NexusTradeseen in 3 videos →https://nexustrade.io ↗

Why it worked

The video taps into a current trend of AI-assisted development, offering a contrarian but relatable perspective on how software engineering is evolving. The creator's personal experience and direct address to the audience make the information engaging and credible.

Summary

A senior software engineer explains that AI models are now so advanced that they can write high-quality code, reducing the need for manual code review. Instead of reading code line-by-line, engineers focus on setting up guardrails and ensuring the AI's output meets expectations.

Structure

1AI models write code, reducing manual review
2Engineers focus on guardrails and quality
3Junior engineers lack experience to guide AI
4Senior engineers trust AI for code generation

Product placement

NexusTrade (ACTS - interface is operated, generates code, visible result comes from it, creator says what it does for them. Removing it would change what happens in the video)

On-screen text

Software engineers are NOT reading code anymore

Original caption

Replying to @Austin with modern AI models, we simply give the model a spec, tell it to follow existing patterns, then let it run. We don’t need to read all that code anymore. ##aiagents##claudecode##swe##cs

More from @starkstechnology

Their other posts in the index, biggest breakout first.

Transcript

Software engineers are NOT reading code anymore. I'm a senior software engineer and I have all of these different AI sessions running at the same time. I have more than five different code sessions, a little cursor session and even a codex session all running simultaneously. The question that I got is do I look at the code if at all? The honest answer is not really. At this point in time, the AI models are so good that they actually implement the code you want to write. Nowadays, I give a cursory glance to make sure that the code isn't doing anything crazy. Then I built guardrails to make sure the code is high quality. This includes unit tests, integration tests, running golden pass to make sure that the outputs are what I expect. Rigorous code review pipeline. I mean, most of the time my AI is just reading the code, it doesn't write the code. Making sure the code actually follows the established patterns. Things like I only really read the code if it's like the AI isn't fixing it within like a single prompt. I do think junior engineers are at a huge disadvantage because they really don't know like what to look for. They didn't struggle to write high quality code. So they don't really know what to look for. For a senior engineer, you're not really reading all the code anymore.