I ve worked on a small side project on and off for about eight months. Some weeks I m excited about it, and other weeks I wonder if I m just refusing to let it go because of the time I ve already put in.

Last month I set a simple rule: if I couldn t get even five people to try it, I d stop. I managed to get three, and two of them gave me useful feedback. I m still not sure if that means I should keep going or if I m just reading a weak signal as a good one.

It made me wonder if don t quit too early and don t hold on too long are both true, and how anyone knows which one applies.

How do you decide when a project deserves more time? Have you ever quit something and later regretted it, or kept going and wished you had stopped sooner?

Last time was caching, and how it can quietly cost more than it saves. Today s question is a step away from my original itch. I started out asking which feature spent my money. If you sell an AI product, the question that matters more is which customer.

On the demo workspace I use for screenshots, two customers are on the same $29 plan. One costs $3.10 a month in AI calls. The other costs $61.20.

We are entering the last quarter of this year, when we can accomplish some improvements for this year.

On Dec 31st, we will check whether the quarter has been successful and whether the whole year was productive in general.

Public launch directly on Product Hunt vs going on Product Hunt a bit later.

I am looking to compare and learn from experiences of others.

PH launch is a must - so question is only about the timing.

Building a product today guarantees nothing. A year ago, shipping something functional was enough to stand out. Today anyone can vibe code an MVP in a weekend. The barrier to building has collapsed which means the barrier to being copied has collapsed alongside it.

So if the product itself is no longer the moat, what is?

Chipmaker Nvidia announced this week it will be investing up to $100 billion (with a B) in OpenAI to help it build out its computing power. This follows OpenAI s move earlier this month to buy $300B in computer power from Oracle.

For me, it was a very simple, yet really insightful thing: enhance what's already working.

We're running a community forum, and our "Changelog" and "Ask the Community" categories performed really well. But after we featured them on the forum's homepage and applied some changes to improve usability, the engagement boosted significantly.

There is so much advice for founders now that a lot of it starts to contradict itself.

Launch fast, but don t launch too early.

Talk to users, but don t build everything they ask for.

Focus on one thing, but keep testing channels.