# A distributor's order-takers start upselling with AI product picks

MSC's customer service reps see Proton's picks for each customer, like items due for reorder and products similar customers buy, and pitch them on the call.

Category: Call prep
Industry: Manufacturing
Difficulty: moderate
Updated: Oct 20, 2026
URL: https://revstackai.co/use-cases/distributor-reps-upsell-with-ai-product-picks

## Problem

Customer service reps were '100% reactive' and only pitched promo items, so upsell chances slipped by on every call.

## How to do this

1. Load your order history into Proton.
2. Turn on reorder reminders and product recommendations.
3. Before you answer a call or email, check the picks on that account.
4. Pitch one item they are due to reorder or one item similar customers buy.
5. Each month, check which picks turned into orders and coach to them.
6. Paste it into Proton, picks show on each account. Or upload an order history export to ChatGPT, Claude, Gemini or Grok and use the prompt.

## Verdict

Order-takers become sellers without memorizing the catalog.

## Tools

- [Proton.ai](https://www.proton.ai/): Reorder alerts and AI product picks for distributors
- [ChatGPT](https://chatgpt.com/): Upload an order history export and ask who is due to reorder
- [Claude](https://claude.ai/): Upload an order history export and ask who is due to reorder
- [Gemini](https://gemini.google.com/): Upload an order history export and ask who is due to reorder
- [Grok](https://grok.com/): Upload an order history export and ask who is due to reorder

## RevStack prompt

Written by RevStack. This is not the sourced result below.

Inputs:
- Order history export (customer, item, date, quantity)
- Customer to focus on, if you want one

Here is our order history export. For each customer, list the items they buy on a regular schedule and are now due to reorder, based on their usual gap between orders.

Then name one product that customers with similar orders buy and this customer does not.

Focus: {{customer}}

Use only the data in the file. If a customer has too few orders to tell, say so. Do not invent products or prices.

Example output, written by RevStack, not a reported result:

Customer, item due to reorder, last order date, usual gap between orders.
One suggested add-on item, with the similar customers who buy it.
Customers with too few orders are marked: not enough data.

## Result

'On average, MSC sees a 13% conversion rate on the recommendations suggested by Proton, with reps generating approximately $15K in upsell/cross-sell revenue per year.' CEO Erik Gershwin: 'we've seen 20x the results that we did with the last [system] we were tinkering with.'

Vendor source. Reported by [MSC (Proton case study)](https://www.proton.ai/case-study/proton-turned-msc-call-center-into-sales-powerhouse).

## Before you start

- Proton (quote only)
- Your order history loaded into Proton
- Reps who talk to customers by phone or email