AI is draining our energy. Human is supposed to focus on limited things at a time. That’s how out brain is wired. Now the expectation is to be able to focus on 10 different things using AI.

Crazy development happening in AI assisted software development. People are creating open-source versions of billion-dollar software. Someone created open-source Adobe suite. It has almost everything from Photoshop to Premier to Adobe acrobat replacements. getartcraft.com/apps

Now that automatic lyrics generation is live, I am trying to add more lyrics display style so you can give even more character to your videos. Excited where Meloviz is going. #buildinpublic #MusicIsLife

You can spend weeks predicting how users will use a product. Then someone uses it in a completely different way. That’s why real usage is so valuable. Users don’t just validate your assumptions. They reveal assumptions you didn’t know you had.

Technical debt isn’t always bad. Sometimes you deliberately take a shortcut because learning quickly matters more than building perfectly. The problem isn’t taking debt. The problem is forgetting you took it.

This weekend, I added automatic transcription and timing on Meloviz using cloudflare workers AI. Chose it for its reliability and cost effective rates. Try it out at meloviz.com #buildinpublic

AI made writing code faster. It didn’t make knowing what to build faster. I think product judgment is becoming more valuable precisely because implementation is getting cheaper.

One underrated advantage of being a technical founder: You can turn an idea into something real very quickly. One underrated disadvantage: You can turn every idea into something real very quickly. The ability to build is powerful. Knowing what to build is even more important

What do you do when AI struggles with little fixes that bothers you? Some part of UI is clipped and no matter how I explain, it just doesn’t get it right.

The fastest product development loop I’ve found is: Build → ship → watch → listen → fix → repeat. Not: Plan → build → polish → build more → finally launch. Perfection before exposure feels productive. Feedback after exposure actually teaches you something.

I’m increasingly interested in software that helps people create rather than just consume things. There’s something different about a tool where the output didn’t exist before you opened it. creative software should make you feel like you made something,

Fixed some bugs in meloviz that caused visualizations to not look as good. Hopefully users will like the little improvements I am making. Next will be automatic lyrics transcription.

Pricing a SaaS is surprisingly uncomfortable. You spend months building something and then have to put a number Too low and you undermine the value. Too high and you scare people away. And the hardest part? Don’t know the right answer until you start charging real customers.

AI hasn’t made software engineering irrelevant. It has made some parts of it incredibly cheap. Which makes the expensive parts more obvious: Understanding the problem. Making good decisions. Knowing what not to build. And taking responsibility for the final result.

One of the best feelings in software: You have an idea. You build it. You deploy it. And then you see an actual person use it. No architecture diagram can compete with that moment.

I am having opus 4.6 moment with opus 5.5. 4.6 is the first time I started relying on generated code. 5.5 is the first time I have utmost faith in the quality it generates. This is insanely good and scary.

Dangerous phase in SaaS where everything starts looking like a feat opportunity. User asks for something → build it. Competitor has something → build it. Eventually you realize you’ve built a product instead of solving a problem. Learning to say “not yet” is a product skill

I used to think the more time you spend planning a product, the better the outcome. Plan enough to avoid obvious mistakes. Then build. You’ll learn more from 1 week of real users interacting with something than 3 months of thinking about what they might want.

Why is it so hard to get the first 100 followers on X ?

A working product isn’t necessarily a good product. You can have: ✓ clean architecture ✓ good UX ✓ fast infrastructure ✓ automated deployment ✓ beautiful UI …and still build something nobody cares about. Distribution and customer feedback are part of the product.