An idea, a carefully crafted prompt, AI agents – and a good night's sleep.
Remember the good old Doodle?
Back when finding a date was simple? Create a poll, share a link, let everyone vote. No registration, no complicated setup, no unnecessary features.
Sunday evening, after dinner, I thought: Why not bring back that simplicity?
And since it's 2026, why not build it myself?
First, the thinking. Then, the building.
I actually invested quite some time in the very first step – developing a detailed initial prompt together with ChatGPT.
Not just describing the functionality, but thinking through the user experience, the architecture, the design principles and an important requirement: no platform lock-in.
The goal was a modern, beautiful, mobile-first application that anyone could use without creating an account.
With that specification ready, I turned to Replit and its modern agent-driven development capabilities.
The AI agents did the coding. I concentrated on product decisions, reviewing results and refining the experience.
Then I went to bed 😴
Monday morning, I continued with a few adjustments, tested the application and released it.
From idea to live product in roughly 12 hours – including a perfectly normal night's sleep.
The result?
👉 meet.rocks 🤘
Free, simple and hopefully as intuitive as the original Doodle.
The bigger story isn't about scheduling meetings
Let's be clear: A date-polling application isn't rocket science. And I'm certainly not claiming to have invented something revolutionary.
But the economics are remarkable.
A few years ago, even a relatively simple web application required considerable development effort. Today, one person with domain knowledge, a well-thought-out specification and AI agents can deliver a working product overnight.
Of course, a functioning application isn't automatically a secure, scalable and sustainable software business. Testing, security, maintenance and operations still matter.
But the barrier between having an idea and turning it into software has practically collapsed.
And that's what fascinates me.
After building considerably more complex applications in investment management and enterprise portfolio steering with AI agents, meet.rocks is a small, almost playful demonstration of the same fundamental shift.
The bottleneck is moving from implementation to imagination, product judgment and distribution.
Or, put differently:
Building software is no longer the hard part. Figuring out what people actually want – and getting them to use it – is.
In the meantime, if you need to organise dinner, a sailing trip or your next meeting, give it a try.
Feedback is very welcome – and sharing is explicitly encouraged 😉
Built with AI. Made for humans.