What Enterprises Can Learn From NTT’s Approach to Customer Insight

Why It Matters

By delivering more accurate intent scores and prioritizing high‑impact interactions, LAMs can boost sales efficiency, lower outreach costs, and enable cross‑industry decision‑making based on unified time‑series data. This signals a shift toward scalable, sequence‑aware AI that enterprises can adopt without the heavy infrastructure of large language models.

Summary

NTT and NTT DOCOMO have introduced a Large Action Model (LAM) that analyzes the sequence of customer actions across digital and physical channels, moving beyond traditional demographic segmentation. The platform structures data using a 4W1H framework and leverages time‑series modeling to predict next actions and rank outreach effectiveness, training on eight NVIDIA A100 GPUs in under a day. In DOCOMO’s telemarketing trials, the LAM doubled order rates for mobile and smart‑life services by delivering timely, intent‑based contacts. The approach is being explored for healthcare and energy applications where timing of events is critical.