Building internal AI platforms: From RAG pipelines to platform capabilities

In this talk, Himanshu Goel examines how teams can move beyond isolated AI experiments and build internal AI platforms that scale across the organization. Using retrieval-augmented generation (RAG) as a reference architecture, the session shows how AI pipelines can be transformed into platform capabilities with clear interfaces, ownership, and guardrails.

The talk covers:
• When AI systems become platform problems
• Designing RAG as a shared platform capability
• Balancing developer experience, governance, and reliability
• Common failure modes when operationalizing AI at scale

Attendees will gain a practical mental model for integrating AI into platform engineering practices without turning platforms into fragile experimentation layers.