Build the MVP Before the Pitch Deck

A founder sat across from me in July with a forty-slide deck, a two-page Notion doc, and a quote from a development agency for 380,000 dirhams and five months. She had been "about to start" for a year.

We built the first version of her product that afternoon. Not a mockup. A working version, with real data, that she could put in front of a customer the next morning.

That's the session. Here's what happens in it, and why I think it's the most useful thing I do for startupers.

The quote is the problem

I ran a digital agency for ten years. I've written that quote. Five months and a six-figure number is a fair price for a team to build a product properly, with a design phase, a backend, testing, and a launch.

It's also the wrong thing to buy when you don't yet know what the product is.

Nobody knows what the product is before it exists. The founder thinks she does. The deck says she does. But the moment a real user touches the thing, half the assumptions collapse and the other half turn out to matter in a different order. Every founder knows this in theory. Very few can afford to learn it in practice, because learning it costs five months and a quote.

Vibe coding changed the price of learning. Not the price of building a proper product. The price of finding out what the product should be.

What we build in a day

The format is simple. One founder, or a founding team of two or three. One idea. A room, or a screen. Six hours.

We spend the first hour not building. I ask the question every dev shop skips because the answer costs them revenue: what is the smallest thing a customer would pay for, or at least use, this week? Not the platform. Not the marketplace with the two-sided network. The one flow. The founder who wanted a full logistics platform left that hour with "a WhatsApp number that takes a delivery order and returns a tracking link." That's a product. The platform can come later, if it's needed at all.

Then we build it, together, with Claude Code and the founder driving. I don't write the product for them. I sit beside them while they describe what they want, and I teach them how to describe it precisely enough that the tool builds the right thing. That distinction matters more than any framework. A founder who has watched their own words turn into working software understands their product in a way no deck can give them.

By mid-afternoon there's something running. Ugly, usually. Missing the login page, always. But real: it takes input, does the thing, and gives an output a customer would recognise.

The last hour is spent doing what nobody does: putting it in front of someone. A friend who runs a shop, a cousin in the target sector, a first customer from the founder's phone contacts. We watch them use it. We write down what breaks.

What changes afterwards

The deck changes. That's the first thing.

A deck that says "we will build" reads as a request for permission. A deck that says "we built this in a day, here's what twelve users did with it, here's what we learned" reads as a report. Investors in Morocco are a small circle, and they've seen a lot of permission requests. Reports are rarer.

The ask changes too. The founder with the 380,000-dirham quote is no longer raising money to find out if the idea works. She knows it works, roughly, for a specific slice of customers. She's raising money to make it work properly, at scale, with a team. Different number, different conversation, much better odds.

And the founder changes. This is the part I care about most. She's no longer someone who describes software to people who build it. She has built it. Badly, quickly, with help, but built it. Every conversation with a future CTO, a future dev shop, a future investor is now a conversation between people who have both touched the thing.

What AI can't do for a founder

I want to be honest about the limits, because the hype around vibe coding does founders no favours.

The tool will not tell you if the market exists. It will build a beautiful solution to a problem nobody has, faster than ever. That's what the first hour is for, and it's why I refuse to skip it.

It will not distribute your product. The hardest problem for a Moroccan startup is rarely the software. It's getting the software in front of the right people, in a market where WhatsApp is the sales channel and trust is built in person. The MVP session gets you to something you can show. Showing it is still your job.

And it will not make the second version easy. The prototype we build in a day is not the product you'll scale on. Somewhere between a hundred and a thousand users, it will need to be rebuilt by people who do that for a living. The difference is that you'll rebuild something you understand, with money raised on evidence, rather than building the wrong thing first with money raised on hope.

The founders I see coming

The ecosystem here is moving. The FORGE at UM6P picked six AI startups for its demo day this year, and what struck me is that they're all in what they called the messy middle: compliance, staffing, restaurant operations, WhatsApp agents. Unglamorous, specific problems in specific sectors. Exactly the kind of thing a founder who knows the sector can prototype in a day and validate in a week.

That's the founder I want in the room. Not the one with the best deck. The one with the deepest knowledge of a boring problem, and a year of "about to start" behind them.

Bring the idea. Leave the quote at home. We'll build it, and then you'll know what to ask for.