Project
RAG assistant foundation
pgvector pipeline and multi-provider LLM fallback chain for Tan.
- AI
Tan, our scoped sales assistant, required a retrieval pipeline grounded in CMS content, provider fallbacks for resilience, and strict guardrails. This showcase covers the architecture we use before enabling the public chat widget.
Outcome
Production-ready RAG ingest hooks and spend controls
Tech stack
- pgvector
- Google Embeddings
- OpenRouter