The Prototype Is Now Free. That's Not the Part You Were Paying For.
Replit's free mode makes demos cost nothing, but auth, data modeling, payments and reliability still cost exactly what they always did. Here's where the free option stops
A founder sent me a Loom last year. Ninety seconds, screen recording, genuine excitement in his voice. He had built the product. Signup screen, dashboard, the core workflow, all of it clickable. He had done it in a weekend without writing code, and he wanted to know what it would cost to "just finish it."
I watched it twice. It was good. It was also about 10% of a product, and the 10% that costs the least.
This week Replit introduced a free mode that lets you generate roughly 30x more without burning credits on everyday work. I haven't used it. I don't need to, because the direction has been obvious for two years and the only thing that changed is the price of the easy part going to zero.
That is genuinely good news. It is also the moment a specific bad assumption gets much more dangerous.
The assumption I want to kill
The assumption is: if I can generate a working prototype for free, I don't need a technical partner.
I understand where it comes from. You raised money. You are watching the burn. Someone quoted you a number to build the thing, and now a tool is offering to do it for nothing. Of course you look at that and think the line moved.
The line did move. It moved sideways, not in your favor. What got cheap is the demo. What is still expensive is everything that turns a demo into a business, and none of it shows up in a Loom.
What I actually get handed
I get called in as interim CTO after these builds hit a wall. It is not once in a while, it is a category of engagement now. The wall is always in the same three places.
Auth. Not the login screen. The login screen is easy and the generators nail it. What breaks is everything that hangs off identity: password resets that actually deliver, sessions that expire correctly, one user who should not be able to read another user's records, an admin who needs to impersonate a customer to debug something, and the day you sell to a company that wants SSO. Most generated apps have a login form and a shared database where any authenticated request can, with a slightly modified ID in the URL, read anyone's data. That is not a bug you notice in a demo. It is a bug you notice in a security questionnaire, or in a lawsuit.
Data modeling. This is the quiet killer. A prototype gets a schema that fits the happy path the founder described. Then reality arrives: a customer has two accounts, an order gets partially refunded, someone changes their email, a record needs to be soft-deleted for compliance instead of actually deleted. Each of those is a change to the shape of your data, and by the time you notice, you have real customer records in the old shape. Migrating live data is the single most expensive category of work I do. It is expensive precisely because it was cheap to skip at the start.
Payments. Stripe's checkout is genuinely easy to drop in. Then you need proration when someone upgrades mid-cycle, dunning when a card fails, what happens to their data during the grace period, refunds that reconcile with your own records, and a webhook handler that behaves correctly when Stripe sends the same event twice, which it will. I have never seen a generated build handle duplicate webhooks. When it goes wrong you either double-charge people or you silently give away product for free, and you find out from a customer, not a dashboard.
Behind all three sits the fourth thing, which is reliability, and the fifth thing, which is a name. Somebody has to be accountable at 2am when the thing is down and your first ten paying customers are the ten people most likely to tell everyone. A generator does not carry a pager.
I built FlipSmrt solo, so I know the ratio
I co-founded FlipSmrt and built it to production by myself, with real MRR. That is the honest test case for this argument, because I am exactly the person who should be able to skip the expensive parts. I know the stack. I have no coordination cost. Nobody is blocking me.
The build that worked in a demo took a fraction of the total time. Everything after was the same list: getting identity right, reshaping data once real usage disagreed with my assumptions, making billing survive edge cases, and then the long, unglamorous tail of monitoring, error handling and the small repairs that keep something running while people pay for it.
That was me moving as fast as one experienced person can move. A founder who cannot read the code does not get a better ratio. They get a worse one, because they cannot see which of the generated decisions are load-bearing and which are throwaway.
So what should you actually do
Use the free tools. Aggressively. I mean it.
Build the prototype yourself before you talk to anyone technical. Show it to twenty potential customers. Find out whether the workflow you imagined is the workflow they want. Use it to close a design partner or to get a term sheet. That is a genuinely great use of a weekend and it is now free, and a founder who arrives with a clickable prototype and ten customer conversations is a much better client than one who arrives with a slide.
Then stop before you take money from anyone through it.
The moment a real customer's data lives in the system, or a real card gets charged, you have crossed from prototype into product, and the cost profile changes completely. Not because the code is bad. Because the requirements changed and nobody updated the build.
My fixed-scope MVP is $14,500 and it is not paying for the screens. The screens are the cheap part now, and I use the same generators you do. It pays for the decisions underneath them: what the data actually looks like, who is allowed to see what, what happens when payment fails, and what happens when it breaks. If you are not sure you are at that line yet, $4,500 buys a two-week fractional CTO sprint where I look at what you have built and tell you honestly what it would take, including the version where the answer is "keep going, you're fine for now."
The prototype being free does not mean you need less help. It means you should stop paying anyone for the part that got free, and start paying attention to the part that never will.
If you have one of these builds sitting in a tab and you want to know how close it is, book thirty minutes. I will tell you where the wall is before you hit it.