Greenhouse 🌱

Inspiration

Everybody has ideas. Almost nobody ships them.

We have all had the moment: you read a complaint online, or hit some annoying problem yourself, and think "someone should build an app for that." Then life happens. The thought evaporates by lunch. Multiply that across millions of people every day and you get an enormous graveyard of good ideas that never got executed.

The gap was never imagination. It was execution. Coming up with the idea, validating that people actually want it, and then sitting down to build the first version. That is three separate walls, and most ideas die at the first one.

We wanted to knock all three down at once. So we built Greenhouse: a place where ideas already backed by real demand get surfaced, and where executing on one is as easy as swiping right.

What it does

Greenhouse reads the open web for what people are actually frustrated about and wishing for. It turns those real signals into buildable app ideas and serves them as a swipeable feed, a bit like TikTok for app ideas.

Every idea on the feed comes with the real quote and the real source that prove someone wants it, so you are never swiping on something we made up. You deny the weak ones and approve the ones you like. The moment you approve, Greenhouse builds a working first version and gives you a live link.

Idea to validated to built, in the time it takes to scroll a feed.

How we built it

The heart of it is a scout: an agent that searches the live web, Reddit, Hacker News, and X, then filters for genuine demand rather than noise. It hunts two things: problems people are complaining about right now, and great ideas from years ago that somehow nobody ever built.

As soon as it finds a grounded idea, it streams it straight to the feed so the page fills up in real time. Each idea carries a short pitch, the evidence behind it, the features a first version needs, and a suggested stack.

When you approve an idea, a coding agent takes over and builds a minimal version of it, then deploys it so you get a real, shareable URL. The whole thing runs on a lightweight backend with a clean, fast front end, and it is deployed so anyone can try it.

Challenges we ran into

The biggest one was trust. The entire idea falls apart if the app invents demand that is not there, so we forced it to quote real people word for word and attach the exact source to every idea. If you do not believe a card, you can check it in one click.

The second was keeping the agent focused. Left alone, it would wander forever, so we gave it strict limits on how long it could search before it had to commit to an answer.

The third was the build step. Getting an agent to reliably produce something that actually runs and deploys took a lot of tuning, plus a simple fallback for when things go sideways.

And finally, making it all feel alive. Ideas arrive, builds run, and cards move from "building" to "built" all at the same time, so we had to keep the screen perfectly in sync with everything happening behind it.

What we learned

The bottleneck in building software is no longer writing code. It is knowing what is worth building, and actually starting.

We also learned that showing your work is a feature, not a footnote. Putting the real quote next to every idea is what makes people trust the feed and enjoy using it. And an agent only feels intelligent when it is on a tight leash and when you can watch it work in real time.

What's next

We want Greenhouse to learn your taste. The more you swipe, the better it should get at surfacing the ideas you, specifically, would want to build. Richer first versions, and a way for teams to grow ideas together, come next.

Want a tightened Devpost-summary version of this too, or is this the right length?

Built With

• typescript