top of page

AI Case Study

Bradesco bank increases customer satisfaction and service efficiency by implementing a virtual agent to aid employees and automate responses

Bradesco bank in Brazil has implemented a conversational agent to interact with customers and employees in Brazilian Portuguese, called Bradesco InteligĂȘncia Artificial (BIA). It can answer both speech and text questions, and is presently capable of handling 94% of all questions asked.

Industry

Financial Services

Banking

Project Overview

BIA was created with IBM Watson and is available on mobile devices (iOS and Android) through the bank's app. BIA is used by 60,000 employees answering questions about products and services. As a virtual agent, BIA uses natural language to manage 30,000 queries every day, and is constantly updated with new information. Within the first nine months of training, BIA was capable of answering questions about 49 products within the Bradesco portfolio with an extremely high level of accuracy.

Reported Results

Bradesco claims that BIA has resulted in:

* Customer satisfaction levels above 85% by reducing time to answer customer question
* Capable of answering both general and complex questions about 62 products and services
* 85% accuracy in responding to contact centre questions, 98% accuracy for written questions, 83% for speech-based questions

Technology

Function

Customer Service

Contact Centre

Background

Bradesco bank is one of the three largest financial services companies in Brazil. It "needed to broaden the scope of customer service, so that all customers could get the answers to their queries as quick and precise as possible. Their plan entailed using a virtual agent to answer tier 1 level customer support questions, easing the Bradesco staff’s burden to dig through databases to find answers".

Benefits

Data

Ongoing data collected from Bradesco customer interactions with the virtual agent, written and spoken

bottom of page