His framing was blunt. Around eighty percent of product data is locked in siloed systems, and around eighty percent of unstructured knowledge sits trapped in emails, documents, and tickets. Generating a design, a requirement, or a test plan is already easy. The hard part is a product that holds together as one system. To make that possible, SPREAD has productized the ontology platform it built over seven years and put agents in charge of it top to bottom.
In the keynote, Philipp covers:
- Why the ontology has to come before the agents
- What SPREAD launched: agent-assisted ingestion, a schema builder for your corporate ontology, and an AI Studio on the ontology layer
- How expertise shifts from operating tools to deciding what good looks like
As Philipp put it on stage: "One operating system. Agents now run it, top to bottom."
Watch the full keynote above, and explore the connected model behind it on the SPREAD platform and in Product Explorer.
Philipp's keynote built on Robert Goebel's case for context engineering as the foundation of industrial AI, and the day closed with the SPREAD and Salesforce announcement on connected vehicle diagnostics. Watch the full CUBE 2026 recap.