Cloud Foundry Summit 2026

Cloud Foundry and BOSH operators depend on a large body of operational knowledge: cf-deployment release notes, BOSH releases, component repositories, upgrade guides, job specs, CVEs, GitHub issues, historical incidents, runbooks, and environment-specific customizations. During upgrades and incidents, the hard part is rarely finding one document; it is knowing which sources are current, trustworthy, relevant to the deployed version, and safe to act on.
This talk shows how to design a Retrieval-Augmented Generation knowledge layer specifically for CF and BOSH operations. We will cover what to index, how to attach useful metadata such as component, release version, stemcell, foundation, date, source type, and trust level, and how to combine keyword and vector retrieval for version-sensitive questions. We will also discuss guardrails for common failure modes, including stale release notes, duplicated docs, missing environment context, and hallucinated security or upgrade claims.
Attendees will leave with a practical reference architecture for making CF/BOSH knowledge more searchable, reusable, auditable, and useful during upgrade planning, CVE analysis, and incident investigation.