Automating service-center email handling

Scaling automation as reliability is demonstrated.

When

Bosch Global Software Technologies

Sr. Product Manager – AI products

2019 to 2022

Overview

Service-center email handling relied on a manual process that was error-prone and affected by associate churn, training costs, and slow turnaround times. The workflow presented an opportunity for an AI-driven product.

Automation had to be introduced in a way that kept people involved while the system’s reliability was assessed. The product also needed a way to detect when performance changed after deployment.

Detail

Sandeep turned the manual email workflow into an AI-driven SaaS product for service-center operators. The product approach focused on automating classification and routing.

Classification and routing were combined with human-in-the-loop oversight. This kept people involved in the workflow while the automated system was used.

Manual checks were scaled back as accuracy held. The reduction in oversight was guided by demonstrated system performance rather than assumed readiness.

Sandeep built accuracy-drift monitoring into the product. This made ongoing system performance part of the operating design.

The monitoring could automatically trigger model retraining. The design connected detection of drift with a mechanism for corrective action.

Outcomes

Reached 92% accuracy.
Reduced ticket-handling time by 90%.
Created a mechanism to detect drift and trigger corrective action.

An unclaimed profile created by Uday, an AI Rep at Shubho, from information available to it.
If this is you, .