When building Zyteghost, I set out for myself a few artificial constraints. The first was time. The first version needed to go out in a week, from ideation to launch. This seemed reasonable to me given what is possible with LLM’s. The other constraint was a revenue system. The initial version needed a way for a user to sign up and pay for the service from day one. These seemed like reasonable requirements. And most importantly, both were geared to serve one purpose: to truly follow the “minimum viable product” methodology. I was going to build an AI-powered newsletter generation service in a week that people could try for free and subscribe to after one week if they wanted to continue receiving daily digests. Simple enough.

Generally, I think the concept of “MVP” has become grossly distorted over time. I’m pretty sure the idea comes from Lean Startup, or somewhere thereabouts, but either way, I’ve heard this term thrown about carelessly countless times in my working life. “MVP” has a nice ring to it. Superficially, people understand the intention of the idea. But rarely do people actually put “MVP” to true practice. Oftentimes, we just can’t help ourselves. The thing always needs that “one more thing”.

And so now we live in a world where software is fast and cheap to build (good is yet to be seen) and we can really stretch the concept of “MVP”. Like, really blow it out. That one more feature, a cleaner navigation bar, brighter color schemes, squared off button shapes, animations, ok just one more feature… When the speed at which we can command software is as fast as it is today, why stop at the minimum? Why not go “medium viable product”?

Even with only a week to give myself, these kinds of questions kept coming up. “Well, I could add a newsletter feature in a few hours. Why not? Maybe podcast transcriptions, too?” What was interesting, though, is that given only a week and a limited token budget, these questions were a lot easier to answer. “Yes, we could add a newsletter feature and podcast transcripts. But the basic RSS parser still doesn’t work completely right and we don’t even have Stripe integrated yet. And I just hit my 5-hour usage limit on Claude and we’re two days away from the end of the week.”

Ultimately, here’s why all these questions about additional features were moot and a true “minimum” was the best choice. Because I shipped the thing, spent $100 on Reddit ads over 5 days, got about 100 click throughs, and yet… ended up with exactly zero users after being live for a week. Not entirely surprising, but still pretty disappointing. The newsletter feature would not have made a difference. Better animations would not have made a difference. Technically, the product is minimally viable. It does what it’s supposed to do. What is still totally unclear though is whether or not anyone wants this. And more importantly, is anyone willing to pay for it. That is what we are testing.

This, to me, is the most incredible part of building software today. A product idea could be totally wrong. Or maybe it needs a slight pivot to work. Or we’re marketing it wrong. Or the problem is one of a hundred different variables including these or a combination of them. Who knows. What is incredible to me, though, is that someone can go from idea to product in such a short amount of time and then start testing that idea out in the wild. The trick now is figuring out what “MVP” really means, especially when the speed of development is so high.