$450K Annual Savings with AI Customer Support Automation

Our client is the international leading supplier of commercial equipment in Europe. With 1000 employees on board, they offer more than 25.000 products and count more than 350.000 global customers in 180 countries so far. The company provides all kinds of equipment for restaurants, bars, hotels, laboratories, and retail food markets.

The Challenge

The company was growing 20+% per year — great news for the company. But the increasing number of customers means growing numbers of customer requests.

The chat support waiting time is 5 hours/day;

The text support resolution time starts from a few hours to days;

The call support resolution time durates from 4 to 10 minutes

They were manually tracking records in CRM from WhatsApp, Web, FB Messenger platforms;  

They have multilingual customer service agents.

Our Solution

Our Solution

Our Solution ☉ Our Solution ☉

We began with a 2-day on-site workshop to deeply understand the company’s goals, daily workflows, and pain points—insights we could never get from calls alone.

From there, we mapped the most common pre- and post-sale inquiries, identified which could be fully automated, and estimated the share a chatbot could handle. We then calculated potential savings and ROI and built a demo. Once approved, we moved into MVP development.

We developed a multilingual WhatsApp chatbot that handles seven languages: English, French, German, Dutch, Polish, Turkish, and Arabiс. The chatbot automates 50% of customer service requests and is integrated with their CRM.

We integrated the chatbot with Salesforce and SAP so that every client would be automatically added to CRM with the right customer information service ticket and agents assigned. 

Custom AI Solutions and Multi-lingual WhatsApp Chatbot

The Results

50% of support queries automated within 12 months

◉ Chatbot deployed in 7 languages

0 sec waiting time for customers

€440,000 in annual savings

266% ROI in the first year

90-day payback period

Previous
Previous

Giving Voice to the FIBA Basketball World Cup Mascot Through AI

Next
Next

60% to Nearly 90% Accuracy in AI Internal Outputs