When I use LLMs heavily, I definitely stop exercising parts of my thinking I used to rely on. That can look like regression. But consider: Stone Age humans spent all their energy just getting food. No capacity left for higher-order thinking. Programming and document creation worked the same way for me - they consumed the mental bandwidth that could have gone elsewhere. Now that AI handles that layer, I can think at a different level. In my case, handing all the coding to Claude has freed up real mental space for product strategy - decisions that actually matter. Maybe the question isn't whether we're thinking deeply, but whether we're thinking deeply about the right things.

I'm taking the opposite approach - managed services all the way, and my monthly infrastructure costs are higher than what's described here.

No regrets. Infrastructure isn't the problem I'm trying to solve. The problem is: who's actually going to pay for this?

Optimizing infrastructure before you have customers is like designing a kitchen before you've written the menu. I launched within 72 hours of starting development and went straight to customer validation. The market feedback started coming in immediately.

Infrastructure costs show up in your bill. The cost of slow customer validation doesn't show up anywhere - until it's too late. That's the number I watch.

Some of this will depend on what experience you’ve got. Someone with lots of experience running Linux servers can probably stand up the sort of thing described in this article in a couple of hours from a starting point of being given the Go application source and a credit card.

Lol, try 20 min at most if taking it slow. (I used to be a linux admin)

Hang on, isn't this more a question of minutes if you know exactly what you're setting up and can throw some Ansible at it? What am I missing?

It doesn’t sound like OP was optimizing anything; it sounds like they just knew how to use that stack, and so are able to get customer validation while also spending very little per month.

Fair point. Stack selection is mostly about what you already know. I chose managed services not because I optimized for it, but because that's the stack I'm comfortable with. That said, my real point was simpler: whatever stack you pick, figure out who's going to pay for it before you spend time on infrastructure decisions.

which approach works better depends on your financial situation and your existing setup. if you have money you can invest, then your approach works. if you have more time than money then invest the time instead. when you have built up your servers over the years, when building a new product, you can also do it quickly because the services you need are already running, and firing up a new database or a new server takes just as long as it takes to set up a managed service. but it doesn't add any cost.

Posting on Indie Hackers has worked best for me so far. Both positive and negative reactions are useful. That said, it's inconsistent - some posts get real discussion, others get nothing. DMs are hard. X and LinkedIn both have weak response rates. The exception on X is when you catch someone tweeting about the exact problem at the right moment - then the conversation opens up naturally. The surprise was the Stripe App review process. The back-and-forth with the reviewer surfaced gaps in my feature set I hadn't noticed. The application process basically became a product review.