Boozy Browsing: Pour Decisions in Web Development · Ndevr, Inc.

Interesting Topics in AI & Web Development: Dead Internet Theory, Vibe Coding & AI Agents

·52 min·20 clips

The opener is messy by design. One host still does not know the topic, so the countdown habits, wine choice, interruptions, and half-jokes all stay in the room before the actual conversation starts. Once AI comes up, the episode gets more useful. The cleanest example is a blog thumbnail: ask too vaguely and the result comes back soft, but asking an LLM to help write the prompt first gets closer. Even then, it still needs a human pass. The host says the output only got about 80 percent of the way there after the generated description was fixed, which makes the whole thing feel like work instead of magic. Coding makes that point sharper. A weak thumbnail is annoying. A weak code prompt can send a project sideways, because the person typing still has to know what they want and recognize when the answer is wrong. Tech debt becomes the thread that holds the segment together. The explanation is blunt: old workflows already left behind extra code, inaccurate code, commented-out code, unused functions, copied-and-changed functions, and features pushed in or ripped out under pressure. AI can move all of that faster. The dry little warning is that every prompt can add more debt, since every generated change may take the project farther from cleanup. The episode keeps returning to a grounded take: AI coding helps, but it does not replace judgment, review, or maintenance. The thumbnail analogy works because it scales. If a small creative prompt needs checking, code needs even tighter direction because the mess hides inside the project. The ending stays loose, with mixed feelings about the cold and Arizona and a joke about checking whether a computer has been taken over. It sounds unfinished in the right way, like people talking through the hype until the boring, useful part is left.

As heard by us

A loose AI coding chat that turns vague prompts and tech debt into a concrete tradeoff.

The episode starts in a loose, drink-in-hand register, then settles into a practical case for being specific with AI coding prompts. The strongest stretch comes when a vague request for a thumbnail is followed through to the extra work it creates: the first pass is only half…

Read the full review in PlayNext →

Why you'd press play

When AI coding starts making more mess than progress, this conversation is the one to hear.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Boozy Browsing: Pour Decisions in Web Development