The simplest version of your product that solves the core customer problem, launched quickly so you learn from real usage.
MVP stands for minimum viable product. It is not a broken or sloppy version but a focused one: it contains only what is needed to solve the most important problem for a specific group of customers. The goal is learning, not impressing: do people use it, do they pay, what do they actually ask for?
Many projects fail because they spend months building features nobody requested. Launching an MVP early reduces cost and risk and gives you real data to decide what to build next. AI tools have made building one much faster, as in the 7-day sprint plan. Practical tip: list every feature you imagine, then delete everything not required to complete the core task once from start to finish. What remains is your MVP, and it will usually be smaller than you expected.