How to build an MVP without coding: a two-week plan with AI

You don’t need a developer to find out that nobody wants your product. Two weeks, a free landing page and the nerve to show strangers something half-finished will do it.

This guide shows how to build an MVP without coding. It starts with a plain definition (smaller than most people think), moves through five types you can choose from and a day-by-day plan for two weeks, and ends with three numbers that tell you if it’s working.

What is an MVP, really?

The letters stand for minimum viable product. Eric Ries popularised the term, and in August 2009 he put it this way: the MVP is “that version of a new product which allows a team to collect the maximum amount of validated learning about customers with the least effort.”

The word that matters there is learning. Ries himself warned that the MVP, “despite the name, is not about creating minimal products.” So a one-page site with a sign-up form counts. So does a shared spreadsheet plus a WhatsApp group, provided it answers the question that scares you most.

He learned it the expensive way. At his company IMVU, the first MVP took six months to reach the market. Another time the team lost two weeks building a feature “that absolutely nobody wanted”. Ries wrote afterwards that a simple AdWords smoke test would have exposed how weak the concept was.

Start from your riskiest assumption, the thing that would sink the idea if it turned out false. What’s the cheapest test for it? Usually not an app.

Five types of MVP you can build without code

Several Italian companies began with something this scrappy (we collected some of their stories in Italian startups that started small). We’ll use one invented example from here on, so the types stay concrete.

Say there’s a third-year economics student in Padua called Davide. Over coffee he keeps hearing first-years complain about failing Statistics. His idea: they’d pay an older student for three tutoring sessions before the exam. He can’t code. Below is what each type of MVP would look like for him, next to a real company that started the same way.

Type	What it means	Davide’s version (invented)	A real example
Smoke test	A page or video describes the product before it exists and asks people to sign up	A one-page site, “Pass Statistics in three sessions”, with a form, shared in course group chats	A three-minute demo video took Dropbox’s beta waiting list from 5,000 to 75,000 people overnight
Concierge	You deliver the service by hand, openly, to a few customers	He matches the first ten students with tutors himself, by phone	In October 2007 Airbnb’s founders hosted the first guests in their own San Francisco home
Wizard of Oz	The customer sees what looks like an automated service; behind it, you do the work manually	The site promises a tutor “within 24 hours” and Davide does the matching in a spreadsheet	In 1999 the founder of Zappos photographed shoes in a Sunnyvale store and bought them at full price only when someone ordered online
No-code app	A working app assembled from visual building blocks	A small booking app built on top of the same spreadsheet	Builders such as Glide or Bubble (see the tools below)
AI-built prototype	You describe the product in plain words and an AI tool writes the code	He asks an AI builder for a booking page and tests it with five friends	A quarter of Y Combinator’s winter 2025 batch had codebases 95% generated by AI

The order matters. Barely any software goes into the first three, and they’re the quickest way to learn something true. Save the real app for when people have shown, with an email, their time or their money, that they want what’s behind the page.

Airbnb is a good reminder of how small the start can be. When the service officially launched at SXSW in March 2008, it got two bookings.

How to build an MVP without coding in two weeks

Two weeks is long enough to get real reactions and short enough that you can’t hide in the building phase. Before day one, do the homework from our guide to validating a startup idea: you need to know who has the problem and what they do about it today.

1. Days 1 and 2: choose what you’re testing. Put your riskiest assumption into one sentence (for Davide: first-years will pay for tutoring before the exam). Then, before a single visitor arrives, write down the result that would convince you.
2. Days 3 and 4: make the page. A headline, three lines on what people get, one form. The privacy notice goes right under the form; more on that below.
3. Days 5 to 7: build one action. Pick the one thing customers come for, say booking a session, and make the smallest version that works. Doing it by hand is fine. So is a no-code tool, or an AI builder if that’s faster for you.
4. Days 8 to 10: go and meet your customers. Reach 50 to 100 people who fit your customer. Group chats work, so do LinkedIn messages, and so does a flyer outside the lecture hall. Then phone at least five of them.
5. Days 11 and 12: look after your first users. Yes, personally, slow and a bit embarrassing as it is. Notice every question they ask and every spot where they get stuck.
6. Days 13 and 14: make the call. Put the results next to the threshold you wrote on day 1. Clear the bar and you carry on. Miss it and you have two choices: change one thing and run another two weeks, or drop the idea without feeling guilty.

Changing direction after a test like this is normal. The name for it is a pivot, and after two weeks it costs a fraction of what it would after six months.

Which no-code and AI tools should you use?

The tool market shifts every few months, so think in categories. These names are just the ones you’ll run into everywhere. We’re not endorsing them, and you can manage without any of them.

• Davide’s landing page would sit happily on Carrd or Framer.
• A Tally or Google Forms form would collect Davide’s sign-ups, and a simple pre-order too.
• Sign-ups can pile up in Airtable or Notion, which look like a spreadsheet and work as a back office for the first months.
• If you want an actual app, Glide and Bubble let you drag components onto a screen and connect them to your data, no code needed.
• Automation. Zapier or Make could add a row for each new form entry and send Davide’s confirmation email.
• Lovable and Bolt are AI app builders. Describe the app in a chat window and they write the code.
• Payments can start as a single Stripe payment link, sent out to take a deposit or a pre-order.

Gusto’s July 2026 report surveyed US founders who started a business in 2025. More than seven in ten Gen Z entrepreneurs had used AI tools to help launch it, against 42% of Baby Boomers. More than half used AI to develop the idea itself. And Gen Z founders were twice as likely as older ones to say they wouldn’t have started at all without it.

Professional startups do the same. In March 2025 Y Combinator’s managing partner Jared Friedman said that a quarter of the winter batch had codebases 95% generated by AI. The same conversation came with a caveat: founders still need to read the code, find the bugs and tell good output from bad. For a throwaway test you can skip that. Once payments or personal data are involved, you can’t.

What can go wrong (and how to avoid it)

Building too much

AI tools make this trap deeper. A feature that takes ten minutes is easy to say yes to, so you say yes twenty times, and every one pushes back the day a stranger finally tries the product. Anything that doesn’t help you test the day-1 assumption goes on a list for later.

Collecting emails without a privacy notice

The moment your form asks for an email, you’re handling personal data under the GDPR. According to the European Commission’s guidance, people must be told at the point of collection, at the least: who you are and how to reach you, why you want the data, the legal basis, how long you’ll keep it, who receives it, whether it leaves the EU, and which rights they have, including the right to withdraw consent. Put it in plain language and link it under the form. Ask only for what you need, which for a waitlist is usually just an email.

Getting locked in

Some platforms keep your work inside them. Bubble’s own manual is blunt: its apps “can only be run on the Bubble platform; there’s no way of exporting your application as code”, although your users’ data can be exported as CSV files. For a quick test that’s no problem. But if the MVP turns into the product, you’ll pay to rebuild it, so find out what you can take with you before you commit.

Which numbers tell you it’s working?

Pick one to three numbers. No more. For Davide they’d look like this.

• Visits that become sign-ups. Of the people who saw the page, how many left their email? Your page builder’s analytics will count the visits.
• Second use. How many people came back for a second session without being chased? It’s the earliest honest sign of traction.
• Money, or a firm commitment. Pre-orders, deposits, a signed letter from a business customer. Compliments don’t count, and neither do likes.

We’re deliberately not giving you target percentages, because they swing so much by sector and channel that a borrowed benchmark would mislead you. Choose your own threshold on day 1 and don’t move it afterwards. Run the loop again and again until customers ask for the product before you’ve offered it. That’s roughly what people mean by product-market fit.

Your checklist before day 1

• Your riskiest assumption, in one sentence
• The MVP type that tests it fastest
• A success threshold, written down before any data arrives
• A landing page with one form and a plain-language privacy notice
• 50 to 100 people who match your customer, and where you’ll reach them
• A note on what each tool lets you export if you leave
• A date in your calendar, 14 days from now, when you’ll decide

Money tight? Our guide to starting a startup with no money covers the costs you can postpone, and if you’re under 35, the previous step on our map lists the grants for young founders in Italy. If the test goes well, the next step on the beginner’s map from zero to a first round is checking whether your company could qualify as an innovative startup in Italy. Our Smart&Start Italia guide covers Invitalia’s programme for that kind of company.