Keyword Search Is Dying. Is Your Catalog Ready for AI Agents? — PayPal
This talk details the evolution of commerce search from keyword-based to intent-based and finally to delegation-based, driven by AI agents. It highlights the limitations of traditional product catalogs designed for SEO and ads, which are not easily readable by AI agents. The talk presents experimental results from PayPal on enriching merchant product data to improve agent recommendations, finding that enrichment consistently helps, especially for thin catalogs. It also emphasizes that over-enriching with unstructured text can lead to agent hallucinations, advocating for quality over quantity in enrichment patterns. The core contribution is the understanding of how to adapt product catalogs and data enrichment strategies to effectively support AI agentic commerce.
More from the Multiscale Indexing for RAG trail
- Grep or Embeddings? Agentic Search Over Company Documents — George He, LlamaIndex
- Distill the LLM, Don't Serve It: Search & Personalization at DoorDash — Raghav Saboo, DoorDash
- Building the Document Context Layer for AI Agents — Jerry Liu, LlamaIndex
- Stop Chunking Like It's 2022 — Yuval Belfer, AI21 Labs