The Fifth Elephant 2026 Annual Conference

The misconception is that an “AI-ready” database only needs a vector index and native RAG, which leads to state fragmentation: juggling object storage, a relational database for metadata, and a vector database.

Modern agentic systems—with potentially millions of agents—demand convergence, not fragmentation. These agents drive bursty, schema-exploding workloads, creating, dropping, and querying millions of tables, generating temporary indexes, and requiring instant, consistent shared state.

Legacy database architectures can’t handle this. The traditional catalog becomes a bottleneck, making metadata operations the new scalability challenge. The database evolves from passive storage to the essential coordination substrate for autonomous behavior.

This talk will reveal the architectural breakthroughs enabling Agentic Scale, underscoring why next-generation databases must be agent-first.