Custom RSS Feeds (alastairrushworth.substack.com)

Blaze.email founder launched an extension of his work—blognerd.app—which combines a 1.3M‑post index, automatic parsing/categorization and text embeddings to let you search, discover and turn curated searches into RSS feeds. Beyond ordinary blog search, blognerd offers embedding‑based “similar posts” and “similar blogs” results (clickable grey pills), direct RSS searching and an editor that turns any content search into a live feed. The headline feature is a visual custom RSS builder (inspired by Yahoo! Pipes): you drag nodes on a canvas, start with one or more Input queries, add keyword filters (include/exclude) and end with a single RSS output, so you can mash together multiple topic sources into one tailored stream. For the AI/ML community this is handy as a lightweight, reproducible pipeline for dataset curation, continual data ingestion and model monitoring: embeddings enable semantic relevance screening and similarity joins, while RSS outputs make integration with tooling and experiments trivial. Index scale and vectorized search imply use of vector DBs and batch pipelines, so this approach supports building focused corpora for fine‑tuning, evaluation or research alerts. Caveats include potential echo chambers and selection bias from curated feeds, so practitioners should validate coverage and bias when using these feeds as training or monitoring inputs.