Chroma guide: An exact ID in the wrong category still isn't the answer
Use three original cards to separate similarity, keyword matching and a strict scope limit.
If a catalog mixes everyday language, short IDs and mandatory categories, compare Chroma's retrieval modes. This is the database at trychroma.com, not the color tool of the same name. The August 18, 2025 Chroma Cloud launch distinguishes semantic similarity, keyword-style search and metadata filtering. They answer different questions: does it look similar, do the words match, and is it in the permitted category? A feature list doesn't establish that a given query selects the correct record.
Draw three fictional exhibition cards: “East, K-08, origami table,” “West, K-08, origami table,” and “East, K-80, origami practice.” For “K-08 in East only,” just the first passes. West has the right ID but wrong scope; the third has similar meaning but the wrong ID. Change the question to “paper craft practice in East” without an ID. Only then can the third be considered, and West must still stay out. No product ranking is assumed; identify which rule excludes each card.
For few records and fixed columns, exact equality and a category filter in a table are enough. Compare further only when both similarity and strict scope matter and the team can retain original records, IDs, categories and review notes. Account for how to label unknown scope, explain a missing match and budget human review. No real documents, deployment, code, accounts, connections or queries.
Content note:This article is based on publicly available information and is intended as a reference for evaluating tools. Features, pricing, and related policies may change at any time; please refer to the latest official information.
Original source:www.trychroma.com