Blog
Notes from the team.
Release notes, deep dives, and the occasional rant about search ranking.
-
2026-06-20
Introducing xrecommend
Why we built an embed-first search engine in 2026 — and how four signals fuse into one ranking.
Read post → -
2026-06-15
Why BGE-M3?
We picked the BGE-M3 model for one reason: it returns dense, sparse, and ColBERT-style multi-vectors in a single forward pass. That's three signals for the price of one.
Read post → -
2026-06-10
RRF, explained in 200 words
Reciprocal Rank Fusion is the simplest way to combine multiple ranked lists. It's also one of the most effective. Here's the math.
Read post →