# COACH: AI coaching for field sales teams that cannot ride along everywhere

Canonical: https://mudpie.ai/companies/coach/
Breadcrumb: [Home](https://mudpie.ai/) / [Companies](https://mudpie.ai/companies/) / [COACH: AI coaching for field sales teams that cannot ride along everywhere](https://mudpie.ai/companies/coach/)
Author: Ali Abouelatta (https://mudpie.ai/authors/ali-abouelatta/)
Published: 2026-09-19
Updated: 2026-09-19
Research type: Company profile
Method: Company and accelerator sources checked 2026-09-19. Product claims are attributed to their sources; this is research, not a hands-on product trial.

COACH is an AI coaching layer for in-person sales teams, where managers cannot ride along to every customer meeting. The buyer decision is whether recording and analyzing field conversations will improve rep behavior faster than adding more managers, ride-alongs, or generic sales training.

## What it does

COACH says reps record meetings from a phone, receive personalized feedback on one skill at a time, and get a daily focus. Managers see which reps need attention, while leadership can query a company-wide “sales brain” built from customer conversations. The homepage lists retail, home improvement, door-to-door, alarm, and roofing teams as target use cases ([COACH homepage](https://www.getcoach.com/); [YC profile](https://www.ycombinator.com/companies/ai-coach)).

| Fact | What the public sources say |
| --- | --- |
| Buyer | Field-sales leaders, managers, and enablement teams |
| Workflow | Record meetings, identify coaching opportunities, deliver feedback, and aggregate winning behaviors |
| Public outcome claim | The YC launch says conversion rates rose 32% on average with top French in-person sales companies |
| Public usage claim | The company says thousands of reps use COACH |
| Founders | Mathieu Perez, Thomas Perez, and Nicolas Sebag |

## Why it fits

The product targets the gap between playbooks and actual selling. The best rep's objection handling, timing, and questions are usually invisible to the rest of the team; a searchable conversation record can turn those moments into coaching material. The feedback loop is also more concrete than a dashboard: one rep, one weakness, one next behavior.

The outcome and adoption numbers are company-reported, not an independent sales experiment ([COACH YC launch](https://www.ycombinator.com/launches/SBP-coach-the-ai-coach-for-in-person-sales-reps)). A buyer should resolve consent and recording rules in every territory, audio quality, language coverage, offline behavior, manager review, retention, redaction, and whether the model can distinguish a good outcome from a good sales process. Pricing was not published in the reviewed sources.

The founders have direct operating context. YC says the Perez brothers previously built Alsace Test, a 44-rep field-sales team that reached €3.5 million in 14 months, while Sebag has built software across hospitality, aeronautics, and healthcare ([YC company profile](https://www.ycombinator.com/companies/ai-coach)). That makes the field-sales wedge credible as a problem choice, not proof of the claimed conversion lift.

Short version: COACH is a strong fit for distributed reps with repeatable, recordable meetings. It is a poor fit where recording is culturally or legally impossible, or where managers need a CRM score rather than behavior-level coaching.

## Sources checked — 2026-09-19

- [YC company profile](https://www.ycombinator.com/companies/ai-coach)
- [COACH homepage](https://www.getcoach.com/)
- [COACH YC launch](https://www.ycombinator.com/launches/SBP-coach-the-ai-coach-for-in-person-sales-reps)

## Cohort context

COACH is listed in Summer 2026. In our 2026-09-18 directory snapshot, 119 of 232 listed companies in that cohort have YC’s primary industry label B2B (51.3%). This is a current-directory comparison, not an original intake count or a performance ranking. [Nine-cohort dataset](https://mudpie.ai/research/yc-cohorts-2026-09-19.json).

## Public website snapshot

Observed 2026-09-19T16:18:38.707Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.

| Signal | Homepage observation |
| --- | --- |
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Not observed in this response |
| Docs/developer link | Not observed in this response |
| Pricing link | Not observed in this response |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |

[Public observations](https://mudpie.ai/research/yc-homepage-links-2026-09-19.json) · [Collection method](https://mudpie.ai/research/yc-homepage-methods/README.md). Missing links here do not establish that a capability or file is absent elsewhere.


## Author disclosure

I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.