Avinash

With AI, you can build almost anything. Knowing what to build is the hard part.

I think the hardest piece has been actually finding what exactly is it that our customers need. Again, not what, need. So what is it that our customer, what is it that we can provide to our customers that gives them a lot of value? I think that has been more harder than actually building it because these days with AI, you can literally build anything.

If you build for investors instead of users, you have already lost.

Yeah, I have a controversial answer to this as well, which is don't build AI tools for or tools in general. If you're optimizing for investors and not talking to users, I think you as a founder have already lost, unfortunately.

We skipped microservices and built Cactus as one Rails monolith.

So instead of building multiple small microservices, writing APIs to connect them all together, the way I'm looking at it is building one big monolithic software using Ruby on Rails, which is again, for most people is unconventional, but I'm optimizing for what we call developer happiness, which means that you end up shipping faster

Don't build for each customer. Build for the pattern across many of them.

when you talk to a lot of users, a lot of customers, which everyone should be doing, you start to see patterns, right? So rather than building for each customer, you group this patterns together and you might come up with a much better implementation that works for a larger audience.

Paul Graham's advice to founders: treat your YC partner like a doctor.

Paul Graham recently told that you should treat your YC partner as a doctor, right? You should go and talk to them about every single problem you have.

Customers have wants. Founders should build for needs.

One thing I want to mention in that aspect is every customer has wants. They want a lot of things, right? But do they really need it? That's what you should go behind. If you start building all the wants, I think you'll end up building.