LATIDIA · Investigación

SmartSalesBench: un punto de referencia multivuelta consciente de la resistencia para evaluar a los agentes de ventas de LLM

arXiv: 2610.12975v1Tipo de anuncio: nuevo Resumen: Presentamos SmartSalesBench, el primer punto de referencia conversacional de varios turnos explícitamente diseñado para evaluar a los agentes de ventas del modelo de lenguaje grande (LLM) contra psicológicamente gr

La noticia

arXiv:2610.12975v1 Announce Type: new Abstract: We present SmartSalesBench, the first multi-turn conversational benchmark explicitly designed to evaluate large language model (LLM) sales agents against psychologically grounded, resistant buyer personas. Existing task-oriented dialogue benchmarks assume cooperative users and measure task completion; SmartSalesBench inverts this assumption, placing agents against personas programmed to resist and challenge across conversations of up to twenty turns. The benchmark includes five agent profiles, twelve buyer personas, fifteen smartphone products across six price tiers, and four controlled experiments with 3 iterations totaling 13,680 sessions. Both sales-agent and buyer persona generations are grounded through retrieval-augmented generation (RAG). We introduce nine evaluation metrics grouped into three tiers: primary outcome metrics