While it's true that you need to begin the iteration process with real customers sooner rather than later, you still have to actually build something for them to use. This is still going to be very hard work.
Yes, don't build features nobody has asked for yet, but do completely deliver on at least one core feature that actually makes people want more. Also, if you're a small team, and you crank out sub-par and not very well understood foundations, you're running the very real risk of having it all fall apart from underneath you when the first thing goes wrong. Again, don't go full k8s if you don't need to, but it's still very important to have a solid grasp on every aspect of your system.
You can't just skip the hard parts and get to the money printer. Or maybe you can, in which case, consider me jealous!