Founders are often told to build something quick and rewrite it later. In practice, the rewrite rarely happens and the shortcuts become the product. There is a better middle ground.
Cut features, not foundations
Be ruthless about scope: one core workflow, one type of user, one way to pay. But invest in the parts that are painful to change later: data model, authentication and deployment.
Choose a boring stack
A mainstream, typed stack lets you hire easily and move fast. Novel technology is a cost you should only pay where it gives you a real advantage.
Release to real users early
An MVP is a learning tool. Put it in front of ten real customers in the first weeks, watch them use it and let their behaviour shape the roadmap.