How do you spin up working POCs over a weekend? I'm using a no-code tool and it's taking me months

Look, if it is your first time building it will not be that fast. I don't think it is good idea to be the fastest. It is better to understand your MVP. What is the minimum features your customers need to start using your app. Build them and deploy, then receive feedback and start improving. Don't spend too much time trying to make your UI super beautiful. I've been there. What i want to say is your customers will care about color of the button the last

Usually, people will just reuse code to get stuff up and running quick. Imagine you have a template with a standard landing page, login/sign up page, pricing, etc then all that's left is building out the actual idea and updating the content on the existing pages.

I also made my own saas template which I'm working on updating: https://github.com/ArbaazMeghani/nextjs-firebase-stripe-tailwind-saas-starter

No code tools often have quirks which mean you can spend a lot of time on very simple features. That probably means it's not worth the effort creating the feature now so reconsider if you need the feature now and if not move on. You may even figure out a simpler/quicker way to create the feature in the future once you become more familiar with the tool.

A lot of experienced developers have resources or ideology/experience that they're able to lean on from spending time learning. As an example for game programming I have UI, Audio, Visuals, frameworks, shaders, etc from my time creating.

As a game programmer I got much faster with my development when an artist asked me if my code always needed to be perfect. Once I broke this thought of perfectionism I started implementing things that just worked.

I also think that it's easy to get stuck on features that don't really matter.

For app development I focus on getting 1 valuable feature done before moving onto the next (coding multiple features at once, and ones that don't provide value to the user can inflate scope and development time greatly)

I also code/create with a get it to work mindset then get feedback from the audience to make it good. If the audience hates what I'm making I'll either try to make it good enough or just dump the project.

Remember the more you do something the faster you become at it. Cleaning a new house takes longer than cleaning your old house because you know the layout and elements of the old one.

What's taking you the longest? Is it everything or just the auth stuff?

And using an API rather than doing cloud setup yourself (which you mentioned) is definitely a lot simpler and makes things faster.

For my own project, it took us 4 months to set up the cloud stuff, and we're a team of 3 co-founders (2 technical).

If you're looking for good AI APIs, then I recommend you check out Evoke since that's our main service.

Currently, we're almost done a stable diffusion API, but more will come in the near future.

You get faster by doing the work. There's no shortcut.

Remember that they are also not just experienced in doing the work itself, but most likely are also experienced in knowing what is essential and what they can skip for the MVP.

I have spent a lot of time over the last 10+ years doing this sort of thing. Do it lots, you'll gain experience.
Specifically for authentication, use a service like auth0 (or any other large, safe service)

As a programmer I built auth/user flows code once, and get to reuse that code forever. Maybe the people you are referencing are doing the same.