Incipit

I completed doctoral coursework in history at the University of Puerto Rico and conducted archival research across eight countries in Latin America. The workflow for handling primary source documents is fundamentally broken. You scan hundreds of pages a day, come home with files named IMG_0047, and lose every connection and hunch the moment you leave the archive. Incipit is an AI-powered research archive that turns messy fieldwork scans into a persistent, searchable, relationship-aware knowledge base. Opus 4.7 reads the actual image of degraded historical documents that defeat traditional OCR and extracts structured metadata with confidence scores. The historian verifies every field before it's committed. Each new document is checked against the full archive for meaningful connections. Research hunches recorded at upload time become standing queries that activate when matching documents appear later. Documents that fall outside the current research context are flagged and preserved in a side collection. Built solo. Not a software engineer. Directed Claude Code to build the tool that should have existed for every historian in every archive. Live at incipit.dev.