you've got the idea and none of the code. that's the normal starting point for most founders, and it's a fine place to start. the danger isn't that you can't build it yourself. the danger is paying to build the wrong, oversized version of it first.
an mvp is one bet, not a small version of everything
the most expensive mistake in mvp builds for non-technical founders is treating the mvp as a shrunk-down finished product. it isn't. it's the smallest thing that tests the single riskiest assumption in your idea. if the risk is "will people pay for this", the mvp is a landing page and a checkout, not a dashboard. if the risk is "can ai do this task well enough", the mvp is that one task working, and nothing else.
write down the one assumption that, if wrong, kills the whole thing. build only what tests it. everything else is a distraction you're paying for.
what to cut first
founders instinctively ask for the things that feel like a real company: user accounts, an admin panel, settings, billing, a polished marketing site. nearly all of it can wait. accounts and billing especially are deep, fiddly work that you should not pay for until people have shown they want the core thing. cut them. add them once the bet pays off.
the tech questions you should actually care about
you don't need to understand the framework. you need to understand three things. who owns the code at the end. what it costs to keep running. and whether another developer could pick it up later without starting over. those three protect you. lock-in, mystery running costs, and a bespoke stack only one person understands are the things that hurt non-technical founders months down the line.
a good answer looks like this: you own it outright, it runs cheaply, and it's built on a standard stack any competent developer can take over.
where ai earns its place
use a model where it removes a real bottleneck, not as decoration. "ai-powered" on the homepage isn't a feature. an assistant that genuinely saves your user ten minutes is. if the ai isn't doing one clearly useful job in the core loop, leave it out of the mvp and add it later.
the running-cost truth
a light mvp, a static site with a form, can run at near zero on a free tier. the moment it has a live backend, a database, a model, that changes to roughly £20 to £45 a month plus the model bill. that's not a reason to avoid it. it's a number you should be given upfront so there are no surprises.
how we scope it
we take the one bet, build the smallest thing that tests it from a proven base so it ships fast, and hand you something you own. when the bet pays off, we build the next layer. you're never locked in, and you're never paying for a feature you haven't yet earned the right to need.
that's mvp builds for non-technical founders done honestly: small, owned, and cheap to run until the idea proves itself.