Anushree Sinha

Fluent language from an agentic IR system can be dangerously deceptive, masking critical errors in planning, retrieval, reasoning, and execution that accumulate over time.

LLM-powered multi-agent architectures are poised to revolutionize video recommendation by enabling precise, explainable, and adaptive recommendations that surpass the limitations of static, single-model systems.