From Ingestion to Agents: How AI Teams Build on Document Intelligence — Adit Abraham, Reducto

Why watch

A technical deep-dive on building document intelligence pipelines at scale, covering lessons from processing billions of PDFs. Explores the evolution from basic OCR to agentic approaches that apply token-level corrections (zero for O), table parsing strategies, using VLMs for long-tail document reading, and how agent harnesses solve complex problems like chart-to-table conversion through code interpreters and repeated self-checks. Targeted at AI engineers building production document processing systems.