Get live statistics and analysis of Emrah's profile on X / Twitter

Whatever you do, do it your way!

PlanetScale folks are very happy right about now that they did not put mysql in the company name

Competition isn’t necessarily a red flag—it's proof someone's already paying for what you're dreaming up

If your MVP takes longer than a month, you're probably gold-plating features nobody asked for

Building lots of apps at the same time is possible with AI but getting attention to them is still the bottleneck. What do you think?

If a SaaS is $20/mo when paid monthly and $200/yr when paid annually, and someone signs up for the annual plan and pays $200, then after 2 months decided to cancel, does it sound fair that they’d get refunded $160? The thinking is, they originally signed up for the annual plan but then broke the contract, so to give them a refund, they are essentially moved to the monthly plan and given back the unused months worth of their upfront prepayment. In fact, it could even go as far as what online streaming companies do, and if they are doing the cancellation after a new month has already started (wrt their signup date), then they would pay for the rest of the third month as well (i.e. no prorating) and get refunded $140. What do you think?

Phew long winded post but certainly appreciate the gist, will add it to the vibing routine TL;DR - tell your coding agent to keep functions small - break up tasks into smaller sub-tasks, AI is good at doing one small thing at a time - go back and ask your agent to evaluate its code, break up large functions, etc

My biggest weakness has been frontend dev.. not just the design but getting anything to work, tying the backend to javascript, making CSS work, mobile support on and on.. AI makes it all 1000x easier, and even it struggles with some stuff but does a far better job than I ever could!

There is a difference between obscurity problem and marketing problem.. If you had barely anyone come to your site to see your app, you are too invisible If you have good traffic but noone is biting, then you have a marketing problem: either attracting the wrong crowd, or built the wrong solution Which problem so you have?

How do you keep hitting that publish button, over and over again, knowing there’s no one on the other side? For me, it comes down to two things: - For writing, that I have something I want to say. I always wished people ahead of me told me things that would benefit me down the line. That almost never happened, I had to figure out most things myself, the hard way. I share so I do what was not done onto me. - For projects, that I have something I want to do. It could be a commodity but I probably have a unique angle to it. Or that noone wants the damn thing, but I feel it is interesting/important and it should be done regardless. Worst case, we can call it “art”. - And that I have fun during the process, and even learn a thing or two. I have worked on many side projects over the years and said lots of things on social media that went nowhere, but I still learned a lot and those learnings always helped me reach higher goals in the end.

Why do you leave projects unfinished?

If a SaaS is $20/mo when paid monthly and $200/yr when paid annually, and someone signs up for the annual plan and pays $200, then after 2 months decided to cancel, does it sound fair that they’d get refunded $160? The thinking is, they originally signed up for the annual plan but then broke the contract, so to give them a refund, they are essentially moved to the monthly plan and given back the unused months worth of their upfront prepayment. In fact, it could even go as far as what online streaming companies do, and if they are doing the cancellation after a new month has already started (wrt their signup date), then they would pay for the rest of the third month as well (i.e. no prorating) and get refunded $140. What do you think?

Competition isn’t necessarily a red flag—it's proof someone's already paying for what you're dreaming up

Building lots of apps at the same time is possible with AI but getting attention to them is still the bottleneck. What do you think?

Whatever you do, do it your way!

How do you keep hitting that publish button, over and over again, knowing there’s no one on the other side? For me, it comes down to two things: - For writing, that I have something I want to say. I always wished people ahead of me told me things that would benefit me down the line. That almost never happened, I had to figure out most things myself, the hard way. I share so I do what was not done onto me. - For projects, that I have something I want to do. It could be a commodity but I probably have a unique angle to it. Or that noone wants the damn thing, but I feel it is interesting/important and it should be done regardless. Worst case, we can call it “art”. - And that I have fun during the process, and even learn a thing or two. I have worked on many side projects over the years and said lots of things on social media that went nowhere, but I still learned a lot and those learnings always helped me reach higher goals in the end.

When will rust enthusiasts get that noone gives a flying rats spaghetti meat balls that an application or a library or anything else is written in rust? Except other rust enthusiasts.. Security, memory safety, speed etc.. yes we all care about those things.. written in rust? Nope.. nada. Not one iota.. But rust has all this built in.. maybe.. I'm not going to take your word for it. Show me benchmarks, audits, any kind of data I'm happy to look at.. Written in rust? OK I'll make sure my tetanus shot is up to date I guess..

AI is good enough to vibe code any app in under an hour... ONLY IF the app has a simple UI with basic/standard UX!

Claude Code subagents are the new game in town. You are to curate separate agents to handle specific tasks.. One for UI, one for backend code, one for frontend code, one for debugging, and so on. A single unified AI is not good at every task because it doesn’t have the specific instructions for every task. The idea is you build up the prompt for each subagent over time and make them better at what they do. It feels a bit too cumbersome and if these things go away eventually, all that work will be for naught. That said, I come up with great directives as we work together and it would be a matter of putting them into a markdown file to improve each subagent.. Perhaps it is not that much work. What do you think? Do you use subagents?

gpt-5 and gpt-5-codex are two different models and apparently gpt-5-codex runs on OpenAI API so having a chatgpt account is NOT sufficient. I have not seen this detail mentioned anywhere, so I’m mentioning it

Whatever happened to vibe coded games with three.js? Everyone got so rich they moved to their own islands?

The real dream job is having no job and working for yourself. What's your dream job?

Being first is good if the incumbent ignores you and/or can't follow you while your fledgling little startup grows and gains strength. Otherwise, distribution is king!