Built in the open
Inkyo
Scores a short-form video's viral potential before you post: one Viral Score, eight dimensions, concrete fixes.
SaaSTesting
Why Inkyo
Creators only learn whether a short-form video works after they post it, when it's too late to change anything. Inkyo closes that gap: paste a TikTok or Instagram link (or upload a file), pick your niche, and get a 0–100 Viral Score, eight dimension scores and a prioritised list of concrete fixes — before you hit publish.
How it was built
A continuously refreshed corpus of viral TikTok and Instagram videos, each analysed with Gemini and embedded for hybrid (dense + sparse) semantic retrieval on pgvector. An n8n pipeline keeps the corpus fresh, Supabase Edge Functions serve scoring and coaching behind Zod contracts, and the Next.js frontend ships in English, German and Spanish — built lean by one person with AI in the loop.
What we learned
A single honest number only earns trust when it explains itself: the eight-dimension breakdown with a concrete fix per weak spot is what turns a score into a to-do list. And a live corpus beats static best practice — what "works" on short-form platforms shifts weekly, so the benchmark has to move with it.
Want something like this for you?
It starts with a conversation: 30 minutes, no strings attached.