AI Digest

A local-first pipeline for a daily AI news digest. It ingests public sources, runs a multi-pass LLM enrichment step, and publishes interactive HTML archives with per-run diagnostics.

Staged design: explicit ingest → enrich → validate → render steps, typed schemas, and a diagnostics waterfall viewable in the browser. Enrichment runs locally on a consumer-grade NVIDIA RTX 4090 via Ollama — no cloud API keys required. Covers eleven editorial categories spanning LLMs, image generation, design/creative AI, typography, and robotics.