Apple details Glyph, a multi-agent system for automating enterprise data catalog tagging and governance

Apple’s Machine Learning Research group describes Glyph, a production system that frames column description generation and column type annotation for data classification as cooperating LLM agents orchestrated as stateful graphs. The Descriptor agent grounds generation in the pipeline source code that produces each column, retrieved on demand from an enterprise GitHub via a reasoning–acting tool loop (active Retrieval-Augmented Generation).

The Tagger assigns labels from a governed 275-leaf Data Classification Ontology by running three complementary strategies in parallel: a description tagger, a line-of-business regex tagger, and a metadata tagger backed by a fine-tuned contrastive encoder over a vector database. The system then fuses their ranked outputs with Reciprocal Rank Fusion (RRF).

To improve retrieval quality, the team fine-tunes a 6-layer MiniLM metadata encoder with an in-batch contrastive objective. On an in-distribution held-out split, the fine-tuned encoder lifts same-tag retrieval from NDCG@10 0.55 to 0.92 and MAP@100 from 0.19 to 0.90 compared to the stock base encoder.

The paper reports end-to-end multi-label tagging quality under a recall-weighted F2 objective across three evaluation groups, an ablation isolating each strategy and the RRF fusion, and the engineering decisions that distinguish Glyph from prior column-type-annotation work and commercial value/regex sensitivity scanners. These include a value-free and code-grounded design, per-tag provenance, and graceful degradation, which together make multi-agent LLM cataloging auditable and operable as a production service.