The asymmetry between where data lives and where systems operate is growing.
80-90% of data created is unstructured or semi-structured, yet most enterprise systems are built around structured datasets. Pipelines shuttle data between ERPs, WMSs, and BI tools while ignoring information tucked away in emails, scanned documents, photos, and phone calls.
Access has been a major constraint, so companies push their trading partners into portals and EDI/API implementations. But most vendors don’t live there, and forcing adoption leads to long implementation cycles and strained relationships.
Advances in OCR and AI models have commoditized data extraction, but the real challenge is coherence. Creating a strong foundation requires going beyond field and table checks. Those confirm the data is well-formed, not that it reflects reality. That requires semantic checks across sources.
It’s easier than ever to vibe-code your way into dashboard nirvana. But operators don’t have a dashboard problem. They need signals they can act on. That starts with the data foundation, not the front end.
