MVP vs Prototype vs Proof of Concept

Proof of concept

Audience: you and someone technical. Fidelity: ugly is fine. Data: fake or a single hard case. Lifetime: throw it away. You need one only when a specific risk can kill the company if it fails. Examples: the Bluetooth device actually streams, the model is accurate enough on your documents, the partner API allows the call you are betting on. You do not need a POC for login, a list of bookings, or Stripe. Those are solved. Building a POC "to be safe" on a standard app is a way to delay talking to users.

Prototype

Audience: potential users and, sometimes, investors. Fidelity: looks real, data is not. Lifetime: keep the lessons, not always the code. Figma, a Lovable or Bolt UI, a Wizard-of-Oz where you do the work behind the screen. The job is a reaction: do they try to use it, do they ask when it is real, do they offer to pay. A prototype that nobody outside your team has touched is a design file. If you already have a clickable AI build, the next decision is whether to harden it or restart, not whether to add more screens. That split is covered in MVP vs prototype.

MVP

Audience: strangers. Fidelity: the one job works, including the boring edges. Data: theirs. Lifetime: this is the product you will keep, so you own the repo. An MVP still cuts almost everything. It does not cut accounts, a working core action, or a payment path if the test is whether they pay. "We'll add login in v2" means you shipped a prototype.

• POC: can we? Throwaway. One technical risk.
• Prototype: do they want it? Clickable. Fake data is allowed.
• MVP: will they use it again? Real auth, real data, one job.

What to build this month

If you have not spoken to ten people who have the problem, you are early for all three. Talk first. If the tech is ordinary and people already asked for the thing, skip the POC and the pretty prototype and scope an MVP. If the tech is ordinary and you are still unsure anyone cares, a prototype is the cheap test. If one engineering unknown blocks the whole idea, a one-week POC is cheaper than a 21-day build aimed at the wrong wall. The cost guide is the right next read once you know which of the three you are actually buying.