If you’re a founder going through a rough phase and think it can’t get worse, it can. There’s no bottom to the founder’s pit of misery:.- Your key hire might quit. - Your investor might back a competitor. - Your cofounder might walk. - Even your spouse might. Stay paranoid.

Almost every submission to a take-home assignment we receive during interviews is now single-shotted through AI. And look, using AI is expected. You’ll be using it on the job too. But if you’re applying for a role, just prompting once and pasting the output won’t cut it. People.

What is he even talking about?

RT @hnshah: @tmaiaroto Without quality, people won’t retain. So that’s a baseline but should be set based on what the minimal bar is for th….

RT @jasonlk: Growth always slows in SaaS if you don't keep adding stuff. Like:.- New product(s).- More enterprise functionality.- New buyer….

At this point all the code generated by Claude in the first or even the second attempt if wrong or very poor quality. We are down to an intern level. However, it’s still way faster this way.

We are in the earliest days of AI. Like the days when first PCs were delivered. Everyone who says never will eat their words someday.

RT @harshmadhusudan: Had this poem by Walter Wintle on my wall growing up. “Thinking”. ‘If you think you are beaten, you are.If you think….

RT @AnchitGupta9: This, right here 👇🏽—is the real reason the MiG-21 earned the "flying coffin" tag. No, it wasn’t the LCA delay. No it wasn….

Claude Code is really lazy when refactoring code. Leaves old code at places and claims it is done. Doesn’t test changes. Poor for even SDE 1 standards.

After a break of few days, resumed again. Asked Claude to increase code coverage. The tests broke, and Claude kept insisting that 73% pass rate is good enough! Persuaded it to fix all the tests. It fixed all of them but gave up on one!.

Dedicated to people who want us to “walk to work” in Bangalore!

Claude Code in Plan mode going strong. Churning a big feature every couple of hours. Was able to build a clean UI with just detailed prompts and repeated tweaks.

Founders often think big shifts are simple: change the market segment, ditch the GTM, bring in a few seasoned leaders, and the team will rally behind it. In reality, it’s rarely that smooth. Changing a company’s DNA is hard. Especially when it’s driven top-down by new hires.

Update: Claude Code in Plan mode did the trick! . Gave it a TRD, it gave me a plan, nudged it like I would have nudged a junior engineer. 1 hour later I have a fully functioning app. I have a functioning UI but not a good UI. I am still finding a way to vibe-design-code a good.

Spent the whole day trying to vibe code a serious app. hour 1: wow it works.hour 3: okay no, let’s do it again.hour 6: what kind of garbage is this.hour 8: fuck it, i’ll just do it myself.hour 10: finally! something works.

A real founder superpower: being able to have candid conversations with people outside the company that completely contradict your market position. Most people won’t speak the truth to a founder as they don’t want to offend or burn a bridge.

I’ll never understand why people reply slowly on whatsapp. Is it a power play? Why risk leaving a bad impression just to seem busy or indifferent? In what situations is this seen positively?.

The most valuable change after stepping away from the founder life is that you can binge a Netflix show, eat at a restaurant, or go on a long drive without worrying about the five urgent things you should be tackling instead.

RT @garrytan: This is great until someone gives your product away for free as a loss leader. Or your worst competitor turns it into a capit….

RT @adilankipalle: Indian startup founders haven't picked up on the "CMO is the new CEO" mandate. While SF's launches are breaking records….