I over-engineered my MVP. Here’s what I learned after my first 50 users.

I just finished a soft launch of Pantr AI with about 50 friends and colleagues. The results were a mix of pride and a very cold bucket of reality.

In the first version, I thought I was being helpful with starting menu that guided people: "Click here to create a mockup," "Click here to combine images." I thought I was making it easy, but since i got a lot of messages “What should i choose?” I concluded people were unsure what they really needed to do, it wasn’t natural even though I had a small explaination in each option.

I thought I was building a better product, but this was getting in the way. People didn't want a "wizard"; they wanted a playground. I realised that my power users—the ones actually using it for professional work felt restricted.

Pivoting to a reference system

I stripped everything back. I moved to an infinite canvas where the only "rule" is the @ reference. If you want a style (or person, or obejct from other image), you @-mention an image on your canvas. No wires, no "logic nodes," and no guided buttons.

The feedback has been night and day… I started to get people sharing with me “results”, and good feedback feeling “free”. It was a small victory eheh.

50 Testers, 4 Actives, 0 Payers

Right now the system is stable, it’s working, and I’m proud of it. I have 4 active users who are using it professionally for their daily workflows and explorations, but they are still using the free credits from the launch, and they are still friends, or close people.

Even with a minimalist UI and a stable build, getting that first "stranger" to pay is very hard. My instagram is full of ads about similar apps, "AI fatigue" is real, and I’m fighting against the tide of random AI generated apps and $20/mo subscriptions.

My gut feeling on pricing

I’m following my gut on a credit-based system. I personally dislike the "SaaS anxiety" on creative tools, that feeling of being billed for a month you didn't use. So I’m sticking with pay-as-you-go credits. But in a world of unlimited subscriptions, I don’t know if this is a stroke of genius or a mistake. VC backed companies, probably are forced to go that direction, and if my gut is right maybe we can capitalize on that?

Launch Promo…

I’m now pushing for my first paying users with a "Founder Offer" (50% off for life) that expires May 1st. It’s a move to see if I can cross the gap from "cool tool my friends use" to "a real business."

I’ve added Pantr AI here on indiehackers, some reddit communites, Betalist and I’m planning a more structured launch on producthunt in a week or so. What’s your thoughts/experience about indexing on product directories?

I’m running some reddit ads to target communities around image generation, just because I found it easy to setup. Results are not looking great, as we don’t see any payment yet. But there’s traffic and some new accounts…

Would love to hear any suggestions on how to move from here.

Thanks for reading