AiMi – Anime search by vibe using hardware-aware Generative AI

Summary: AiMi uses a hardware-aware retrieval-augmented generation (RAG) engine with Generative AI to find anime based on descriptive "vibes" instead of tags. It supports dual-mode optimization for CPU or GPU and includes a Viral Receipt Generator that creates shareable watchlist receipts.

What it does

AiMi searches over 8,000 anime titles from 1917 to 2025 using HyDE-based embeddings, auto-optimizing for CPU or GPU performance. It also generates aesthetic receipts from user watchlists.

Who it's for

Users seeking anime recommendations beyond traditional tag-based search and those wanting to share watchlists visually.

Why it matters

It improves recommendation relevance by understanding anime "vibes" and runs efficiently on standard CPUs without expensive hardware.