Show HN: We put voice agent on our website, learned retrieval isn't bottleneck

Category: infrastructure

Tags: semantic-search, ai-agents, real-time

Score: 7.5/10 (Innovation: 7, Technical: 8, Documentation: 8, Utility: 7)

Moss is a sub-10 ms semantic search runtime designed for conversational AI agents, offering hybrid retrieval with built-in embeddings and metadata filtering that runs in-process to eliminate network latency. Its innovative approach of embedding search directly into the application enables real-time voice and copilot experiences that feel alive.

Target audience: AI engineers, backend devs, voice agent developers

Repository: https://www.moss.dev/blog/founding-agent · Python · BSD-2-Clause · 420 stars

View on Hacker News