Hi Product Hunt 👋 I'm Abdallah Albliwi, building ScentHash as a solo maker from jordan. The idea started from a simple observation: everything digital has a format — images (JPEG), audio (MP3), video (MP4), documents (PDF). But scent has none. There's no common way to store, compare, or exchange a smell as data. ScentHash is my attempt at that missing layer. You type a scent in words, and a live, deterministic AI encoder (Llama-3.3-70B via Groq) maps it into: • a base-note recipe (from a 40-note experimental palette) • a 40-dimensional fingerprint • a SHA-256 identity + version Same words → same result, every time. You can also compare scents by similarity, and there's an experimental "lab" view that visualizes the whole pipeline. I want to be upfront (it matters to me): this is v0.1 and honest about its limits. The 40-note model is an experimental representation — not a claim that every smell reduces to 40 notes. The sensor/lab data is clearly labeled as simulated (no physical scent hardware yet). What IS real: the AI encoder, the fingerprint, the hashing, and the similarity engine — all live today. I'd genuinely value your honest feedback — on the concept, the encoder quality, or where you'd take a "digital layer for scent" next. Try it: https://scenthash.com Thanks for taking a look 🙏

ScentHash

A digital representation layer for scent

A digital representation layer for scent