Everyone talks about embeddings, chunking, and vector similarity, but is it only suitable for prototyping? PII leaks into responses, licensing violations stem from company data or internet knowledge, content that should not be shown to specific users, and vector databases that incur hosting costs, reminiscent of 2015. This session cuts through the noise to discuss the practical architecture decisions that make or break RAG applications in production.

The agenda will include:
- Resilient & cost-effective way to build RAG and prepare for growth
- Guardrails and security aspects
- Where to take trade-offs and avoid big surprises

Come for the technical insights, stay for an honest perspective.