Unstructured enterprise data is leaving AI without vital context

Businessinsider: Manish Sood, CEO and founder of Reltio, writes that roughly 80% of enterprise information — including emails, contracts, call transcripts, PDFs, and clinical notes — is unstructured and typically excluded from the data layers that feed AI.

The article says models can read unstructured text but still need clean, connected, trusted context; without it AI systems can hallucinate, miss nuance, or make decisions in a vacuum. It describes meaning-based search and retrieval-augmented generation (RAG) as ways for agents to pull relevant passages in the moment, and explains that Reltio has built technology to link unstructured signals into an intelligent data graph that produces governed, 360-degree profiles tied to customers, products, suppliers, and other entities.

Sood argues enterprises already have the data they need and that the change required is connecting, governing, and reusing existing unstructured information so AI has timely context. The piece says the tools to extract value are available now and that acting to unite structured and unstructured data is the path to improving AI outcomes.

Key Topics

Business, Reltio, Manish Sood, Unstructured Data, Retrieval-augmented Generation, Meaning-based Search