I built Zunder – a tool to enable people to learn more about AI slop and how to distinguish AI from real pictures and videos.
Why Zunder
In recent weeks I was thinking about more and more people can't really distinguish AI slop from actual real photographs and videos. Zunder is about educating people to help distinguish AI from real. Give it a try and tell me what you think. I'd love to get some feedback. Still WIP but I think it is a vital idea.
App: Zunder
Key features
✅ Pattern Library Learn from curated examples of common AI tells—hands with too many fingers, lighting that doesn't make sense, reflections that break physics. Each pattern shows you what to look for.
✅ Community Insights See what others noticed about each image. Different people spot different things, and learning from multiple perspectives sharpens your own detection skills.
✅ Browse & Learn No tests, no scores, no pressure. Explore cases at your own pace, see the community's analysis, and build your skills naturally.
How I built Zunder
Found several subreddits discussing images and videos that might be AI. Notable subs are: IsThisAI and RealOrAI. I learned a lot while browsing discussions and I thought to make this more accessible for everyone to learn. I fired up CLaude Code to plan a MVP. After a long planning session (which burned most of my token credits, he) I had a prompt that I was using to feed V0.dev.
It reliably built a first MVP skeleton and pushed that to Github. With Cursor I pulled to remote Repository and gave Claude Code instrucitons to get familiar with this project. Then I implemented feature that we had plannend together. Bought domain, published. Then I built a curation tool to scrape the Reddit data. Automatic scraping is a huge pain, so that is the MVP process. Curation Tool. I think the manual curation only helps to maintain a high quality of posts. It's not about getting as many examples as possible. It is about helping people to really spot the differences.