AI’s Impact on Data Silos and Knowledge Hubs

A recent KMWorld article, “AI’s Impact on Data Silos and Knowledge Hubs,” makes an important point for enterprises: AI is not a single technology, and it is not a magic wand for eliminating silos. The real opportunity comes from combining language models, vector search, agents, and knowledge graphs into a more intelligent knowledge architecture.

For Franz Inc. CEO Dr. Jans Aasman, knowledge graphs are central to this shift. Semantic knowledge graphs provide a standards-based way to connect information across relational databases, document stores, NoSQL systems, and data lakes. As Aasman explains in the article, modern graph approaches can use language models to help organizations “list all the repositories you have,” extract their schemas, and turn them into an enterprise ontology.

Dr. Aasman Stated, “A knowledge graph gives the enterprise a semantic map of its data—so AI can understand not just where information lives, but what it means.”

This is the key difference between simply indexing content and creating a true knowledge hub. Vector search is useful for finding related content, but it does not replace the need for structure, meaning, governance, and explainability. Aasman describes a system that can take the schema of PostgreSQL, MongoDB, or Parquet and turn it into an ontology, allowing users to query across these systems through a language model connected to an enterprise semantic layer.

 “The goal is not to move every piece of data into one place. The goal is to make distributed data understandable, connected, and queryable.” Noted Dr. Aasman

This also changes what natural-language access can mean. Instead of asking one question and hoping for the best, an AI agent can decompose the question, inspect the enterprise ontology, determine which databases are needed, write the appropriate SQL, Parquet, DuckDB, or graph queries, and perform multiple steps to arrive at a trusted answer.

Dr. Aasman concluded. “The future of enterprise AI is not just chat over documents. It is an AI agent reasoning over a semantic model of the enterprise.”

AI’s impact on data silos and knowledge hubs is still emerging. But one thing is already clear: enterprises that want reliable, governed, and explainable AI need more than embeddings and prompts. They need a semantic foundation that connects data, context, rules, provenance, and meaning.