Before landing on the camera-first approach, I looked at building yet another manual-entry app with a nicer UI — better forms, smarter defaults, quicker taps. But that's the same problem everyone else has already tried to solve, and it still leaves the core friction in place: you have to stop, open the app, and type things in. I also considered a barcode/manual-lookup approach for species ID, but that still requires you to know what you're looking at. The camera-first bet was the harder build (species ID and test strip reading both took real work to get right), but it's the only approach that actually removes the friction instead of just softening it.