The minimum feature set that delivers customer value.
In this approach, you start with a vision of your solution which you keep at the back of your mind. Then you use customer interviews to prioritize the most critical must-have features and scale down your solution into a minimum feature set that still delivers enough value to customers.
People complain that the iPhone didn't have basic features like copy-paste or multitasking when it first launched. Yet, that didn't keep people from lining up for hours to get their hands on one. The first launch of the iPhone aimed to solve one problem and one problem alone that was unifying a phone, an MP3 player, and a web browser in a single device. So you didn't have to carry three devices. Everything else was nice to have and could be deferred to later, even though people complained it didn't keep the masses from buying and loving the first generation iPhone. If your first version gets the basic right, there is always time to add other nice to have later.
The trap that most people sometimes fall into with this approach though is that they get into traditional product roadmap thinking they take their final vision of a solution and chunk it up into a set of releases that are built at some predictable intervals. The problem with this approach is that it presupposes that you know what to build beyond the first MVP which is almost never true.
The MVP lets you start having a meaningful conversation with customers about your solution and you discover what to build next through these conversations through the concept of customer pull. Rather than pushing features out to them, they should be ideally be pulled from customers.
This isn't true just for your MVP release but for every subsequent product release that you also make.
There is nothing wrong with having a big vision but keep it focused on delivering value to customers, not delivering your solution to customers, which are two very different things. Go big on vision but small on solution (product roadmap).
Stripping down your first release to its core is a commonly used tactic for reducing the scope and timeline for a product.