We Reviewed a Landing Page and Accidentally Built a SaaS Product

This one wasn't planned.

I'm 14 days into a 60-day challenge to hit $20K in revenue using an AI agent team to build and launch products. The "Seinfeld crew" — Kramer (ops), Newman (QA), Puddy (research), Elaine (content), George (sales). Each one handles a lane. I coordinate. We ship.

Yesterday the target product was NoShowNinja — an AI-powered no-show reduction tool for healthcare, dental, and salon practices. Missed appointments cost these businesses billions a year in lost revenue. We're building the fix.

Kramer built the landing page. Newman reviewed it and caught three real issues: the contact form was a mailto: link instead of a real capture form, the GA4 tracking ID was still set to G-XXXXXXXXXX, and there was no favicon. Good catches. Newman's thorough.

Then I did my own review.

And that's where things went sideways. In the best possible way.

The Moment

I was doing what every founder does when they're reviewing a landing page: opening it in a browser, tabbing through sections, mentally flagging things. This headline is a little weak. This button copy is off. Is this placeholder text still in here?

It takes time. It's tedious. And critically — you can't easily share it. When you're done, you either write up your notes in a doc (no visual context), or you take screenshots and annotate them manually (slow and ugly), or you just call someone and walk them through it live (impossible to scale).

I sat there doing this old-fashioned review and I thought: why is this still a manual process?

AI can read HTML. AI can identify copy issues, placeholder text, tone inconsistencies, broken elements. We already have tools that take screenshots of pages. Why isn't someone combining those two things — AI detection plus visual overlay — into a single audit tool?

I stopped reviewing NoshowNinja. I started building ReviewLens.

What ReviewLens Does

The concept is straightforward: you give it a URL, it does everything else.

1. Capture — Screenshots the live page
2. Scan — AI reads the full HTML and analyzes it: copy quality, tone consistency, placeholder detection, broken elements, CTA clarity, SEO basics
3. Pin — Issues get mapped back to their visual location on the page as color-coded overlay pins
   • 🔴 Red = critical (broken functionality, missing integrations, major copy errors)
   • 🟡 Yellow = warning (weak copy, inconsistent tone, questionable UX choices)
   • 🔵 Blue = info (suggestions, improvements, SEO opportunities)
4. Report — Full PDF export with all findings, severity ratings, and recommendations

The pins are clickable. Click a pin, see the issue description and suggested fix. It's the kind of review experience that used to require hiring a consultant and scheduling a call.

I had a working prototype running by end of day. Live demo on a Supabase endpoint. Launch Ready Tool (launchreadytool.com) was born.

The Market (According to Puddy)

Once the prototype was live, I put Puddy on competitive research. The headline finding:

Nobody is doing this.

Not exactly, anyway. There are three camps of tools in this space:

Camp 1: Manual annotation tools
BugHerd, Marker.io, Pastel, Usersnap. These are good products. They let teams pin issues on live websites and collaborate on fixes. The problem: humans have to find the issues first. The AI does nothing. You still need someone to look at every inch of the page and decide what's wrong.

Camp 2: AI copy auditors
Tools that will analyze your copy and spit out recommendations. Some of these are solid. But the output is a text report — a list of findings with no visual context. You read "the headline on the hero section lacks urgency" and then you have to go find the hero section yourself.

Camp 3: General AI website analyzers
Broader tools that evaluate SEO, performance, accessibility. Not focused on copy and UX quality. Not visual. Not oriented toward the "I need to QA this landing page before launch" use case.

Then there's the empty quadrant: automated AI detection + visual pin overlay. Nobody's there. ReviewLens is.

The market opportunity scoped: companies running any kind of web presence — agencies, SaaS companies, ecommerce brands, solo founders — all of whom launch pages and have no fast, visual way to audit them before (or after) they go live.

What This Is Actually About

ReviewLens might become a product. It might become a feature inside something else. It might get parked and revisited in 30 days when NoshowNinja is generating revenue. I don't know yet.

But the story matters because it illustrates something I keep seeing in this challenge: AI agents compound.

I didn't sit down and say "what's a good SaaS idea?" I sat down to build a specific thing. The agents did their work. I did my work. And the process of doing the work surfaced the next idea. The QA step revealed the gap. The gap became a prototype in one afternoon.

This is what an AI venture studio actually looks like in practice. Not a person in front of ChatGPT brainstorming business ideas. A system running, producing outputs, and those outputs triggering the next thing. The ideas don't come from a whiteboard. They come from the work.

What's Next

NoShowNinja is still the priority — we're finishing the backend and prepping for outreach this week.

ReviewLens is parked as a working prototype with a live demo. If there's inbound interest from this post, I'll accelerate it. If not, it goes on the roadmap for later.

Either way — it exists. It works. It took one afternoon.

Day 14 of 60. Two products. One planned.

---

Follow along with the full 60-day $20K challenge at growthmindset.ai or subscribe below for weekly updates.