A generative-AI sales coach is running inside a nationwide auto retail network, Michigan State reports
MSU marketing professor Siqi Pei built a generative-AI coaching system deployed with a nationwide automotive retail network - with customer permission it records sales conversations, summarizes needs and concerns, and generates follow-up recommendations. The researcher's framing is explicit - a hidden coach supporting the salesperson, not replacing them. Partner identity, scale, and outcomes are not disclosed.
Source: MSUToday (Michigan State University's own publication)·Entered the record: 2026-08-05·Published: 2026-08-06
What would prove this wrong: This signal weakens if the underlying study shows simulated rather than live use, if the network confirms only a limited test, if use is discontinued, or if no recurring coaching-review work appears.
Associates briefed T+0. Subscribers T+7. This is the public record.
What the record shows
Fetched from Michigan State University's own newsroom by this desk on 2026-08-06. In MSU's account, the system records customer conversations with permission, analyzes each interaction, summarizes the customer's needs, preferences, and concerns, and generates recommendations for follow-up. Pei's own words: "The AI doesn't replace the salesperson... It works as a 'hidden coach.'"
The claimed mechanics are concrete: price questions reframed as total cost of ownership, range concerns reframed around commuting and charging, feature hesitation reframed around safety and family use. MSU states the design could extend to other high-stakes customer-facing training.
Why it matters for people
Routine conversation summary, needs extraction, and scripted coaching are exactly the tasks under pressure in sales roles. What gains weight in the same motion: privacy handling, judgment about customer context, review of machine recommendations, human coaching, and accountable communication. This is the pattern worth learning early - the task list inside the job is being resorted, and the people who understand the new sort first have the advantage.
What this does not show
This is a university's account of its own researcher's work - the partner, user count, duration, sales outcomes, and any job reductions are not disclosed. No dealership-level confirmation exists in this record.
Watch next
The underlying study, when published - live deployment or simulation is the decisive fact. Any disclosure of the retail partner. Whether similar coaching systems appear in other sales-heavy industries.