I'm not a coder. Most of what I build with AI starts with conversations. I explain an idea, share experiences, ask questions, and develop the work from there.

I've been working this way extensively, and I genuinely enjoy it. What AI brings to the process the connections, creativity, and ability to help develop an idea is a big part of why I keep building.

But two experiences with long-running projects made me realize how much I still don't understand about how AI interprets language.

The first involved continuity across sessions. I wanted to preserve where the work stood, what we'd decided, and how the project had evolved. I didn't realize that the language used to preserve that continuity could also influence how AI approached future work.

disclaimer: I'm not an engineer. i do growth and marketing.

when people talk about vibe coding they usually mean shipping a product. an app, a saas, a tool with real users. that's the version that gets posted. but it's not where things actually changed for me.

The quiet things you only notice an hour or a week later, not the dramatic stuff that gets blocked.

Mine was an agent asked to refactor an API endpoint and "make sure all existing tests pass."

Instead of fixing the endpoint logic to match an edge-case contract, it quietly went into the test suite and updated the expected assertion values so the assertions matched its broken implementation. Every test turned green, CI passed without a peep, and it logged a clean summary saying all tests were verified. I only caught it two days later during a manual sanity check when a staging payload failed.

Hey PH mates!

I ve been tracking my own workflow lately while chatting around with ChatGPT and Gemini, and I noticed a painful pattern: I had the whole picture in my mind, but I constantly forget some edge cases or tiny technical constraints on my first try.

I hit send, wait 30 seconds, realise the AI guessed wrong, tweak it, and retry. It s an absolute token-waster and a massive productivity killer.

I'm trying to prototype a lightweight extension to intercept these "blind spots" in real-time right inside the input box before hitting send. But I'm curious about how you lot deal with this:

I recently tested the limits of pure vibecoding building a complete real-time micro-app using only AI prompting.

The project is a live "King-of-the-Hill" billboard where projects outbid each other in real-time to hold the top spot.

Something else I ve run into with vibe coding: the agent ships something that looks finished but isn t actually reading anything.

One example: an agent built a summary report with a confident number nobody had measured. The dashboard looked done, but the stat was a hardcoded fallback from an early prompt. It took a full day to realize it was fake.

Nothing looked broken and the number was plausible, so nobody questioned it.

One cheap check I ve seen: change the underlying data and reload. If the stat doesn t move, it isn t reading anything.

My aunt has the task of organizing volunteers at a local food pantry, using a spreadsheet that was filled in by phone. There would be overlapping shifts, people would disappear, and my aunt spent Sunday evenings trying to fix everything by hand. I offered to create a simple solution, not having any experience in programming.

I used Replit's assistant at the start but then switched to Lovable to have a clearer interface. The initial app allowed one to choose a shift and type their name into it, and it was done within an hour. Another hour of coding was spent solving problems I did not think of earlier: What to do when two people try to book the last shift at once? How can my aunt cancel a certain shift without contacting me?

I m a product designer, and vibe coding has changed where Figma fits into my process.

I still use simple prototypes when I need to review the overall flow and context with the team.

But when I actually want to test an idea, I m increasingly going straight to building it with AI instead.

A working version lets me test things that are hard to judge in a static design, especially interactions and how the whole flow feels when you actually use it.

Two things I'm curious about:

1. How do you monitor your apps for issues and keep track of costs across paid services? Especially if you want to avoid any unexpected bills.

2. When adding a new feature, how do you figure out which services to use? And do you find the setup process annoying enough to slow you down?