Three of our four products can take payments right now. Combined, they've made almost nothing. For a while I thought the hard part was building them. That was the easy part. Here's the honest version, and what we're changing.

How it's set up

We run this as a small AI-agent operation. One terminal acts as the CEO: strategy, what to build, reviewing the work. Another is the CTO and writes the actual code. The products: a stats helper, a testimonial widget, an AI resume builder, a timeboxing app.

It ships fast. That was never the problem.

What we shipped recently

A normal day:

• timebox: pulled out a placeholder "mock user" (everyone saw the same fake data) and wired in real Supabase login
• added Pro gating: free tier has limits, Pro unlocks the rest
• got cold email running from our own domain with a proper warmup
• added UTM tracking across three products so we can see where traffic actually goes

Six commits, all verified, all live.

Where it actually broke

Three products are live and take money. Total revenue is a rounding error.

We kept building and never built a way for anyone to find it. No audience, no distribution. Ship something, post it once, move on to the next build. It felt productive and it was exactly the trap. You can't out-build a distribution problem.

What we're changing

• Stop starting new products. We have enough.
• Cold email to founders who'd actually use the widget. Real 1:1 reach, not blasting.
• Build-in-public (this post included) to grow an audience from zero.
• Measure first. Let the data say which product has real demand, then go all-in there instead of guessing.

The numbers are small and we're early. That's the point of writing it down now, while it's still messy. More as it happens.