Just listened to a podcast about indie AI devs on my commute, and it got me thinking about something I keep seeing advise founders on. There's this gap between what costs money to build and what costs money to sell, and honestly, I think a lot of people building solo or in tiny teams don't account for it properly.
I was speaking with one of my portfolio founders last week. Beautiful MVP. Real technical chops. They'd spent about four grand on compute, some API calls, a bit of cloud hosting. Clean work. But when I asked what they'd spent on actually getting it in front of potential customers, the answer was basically "nothing yet, I'm waiting until it's ready." That's where things fell apart in my head.
In my view, the ready moment never actually arrives on its own. You ship something that runs. That's not the same as something that sells. And the stuff between those two points? That has a price tag most people don't budget for.
Let me break down what I mean. When you're demoing, you're often not running on your actual infrastructure. You're spinning up a Hugging Face Space, or a Replit, or you're running it locally and screen-sharing. That's free or nearly free. But it's also not scalable, it doesn't handle edge cases the same way production would, and it always, always breaks during the demo because that's the universe's standing law.
So you either spend time (which has an opportunity cost) or money (which doesn't) to fix that. Most indie builders choose time, which makes sense, but that time compounds. I've watched founders spend three weeks perfecting a demo experience that will be seen by maybe five people. The ROI math is brutal.
Then there's the hidden layer. To sell anything, you need to prove it works to people who aren't you. That means documentation. That means a landing page that doesn't look like it was made in 2019. That means maybe some instructional content, at least a decent README. For some things, you need a simple dashboard or API endpoint that doesn't ask the customer to run code in the terminal. To be fair, none of that is complicated, but it all takes time and some of it costs money.
I advised someone last month who'd built something genuinely clever. Multi-agent workflow toolkit. They'd sunk about two thousand pounds into development. Solid. But when they wanted to start outreach, they realized they didn't have a consistent brand, no clear pricing model, no documentation that was separate from their personal notes. So they either had to spend six weeks building all that, or pay someone else to do it. They went the time route. It took seven weeks, actually. The compute costs were paid upfront, but the go-to-market costs were all in labor, and that labor wasn't being billed to anyone.
Here's the thing that bothers me about the narratives we tell ourselves. We talk about the cost of building. GPUs aren't cheap, API calls add up, and if you're seriously iterating you might spend real money. That's all true and worth thinking about. But the invisible budget is the distance between demo-ready and market-ready.
Demo-ready means it works when you control the inputs and the environment. Market-ready means a stranger can use it without your help, and it doesn't explode. Getting from A to B has a cost. Sometimes it's cheap. Sometimes it's not.
I think what would help is if people actually tracked both lines separately when they're building. Line one: what did it cost to make this thing work? Line two: what does it cost to prove it works to someone else? And then a third line, which nobody wants to talk about: how much runway do I have left after both of those are paid for, and how long do I have until I need to earn revenue?
Because here's the candid bit. I've seen a lot of impressive technical work die because the person who built it ran out of patience or money before they got it into the hands of people willing to pay. Not because the thing wasn't good. Because the accounting was off.
One of the founders I work with is being smart about this now. They built a prompt pack. Actual development time and cost was minimal, maybe eight hundred pounds total. But they spent another thousand on a landing page, proper packaging, some simple analytics, and a system to handle refunds and customer support. That second thousand isn't glamorous. It doesn't feel like building. But they've been profitable for three months, and the people who used the first version are the reason the second version exists.
So. If you're tinkering with something right now, maybe sit down with a spreadsheet. What's going in? Compute, time at your freelance rate, any services. What does it cost to get it in front of ten people who might buy it? That's the number I'd watch. Not the development cost alone. The development cost plus the cost of proof.
Would love to hear if anyone else is tracking this separately, or if you've hit that wall where you realized the demo wasn't enough.