Bradesco is one of Brazil’s largest banks, with over 5,200 branches.

There’s even one inside a boat on the Amazon. When branch employees had questions about products or services, they called a central office, but there was often a long wait for answers. This meant the client was also left waiting, and as one manager put it, “No one likes to wait.”

In a business as competitive as banking, if your customers don’t have a great experience, they may not be your customers for long. So Bradesco started looking for a way to increase the speed of service and also improve the level of personalization for each client. That’s when they turned to IBM and Watson.

Watson is AI from IBM that seamlessly embeds into your workflows while integrating with the leading platforms and tools enterprises already use. Putting AI at your employees’ fingertips when they need it — and where they need it — means empowering your teams to focus on what they do best.

For Developers & IT: Build innovative cognitive apps with Watson on IBM Cloud.
Try IBM Cloud platform for free →

Create personalized customer experiences by easily connecting applications and data across multiple clouds.
Integrate cloud apps with ease →

Watson Products: Browse the Watson suite of APIs, pre-built applications, and AI lifecycle management tools.
View all Watson products →

Synthesize big data to find the most efficient ways to use your resources.
Connect with a Watson rep →

The right partner, right from the start

“We were IBM’s first customer in Brazil, so there was a very trusting relationship from the start,” said Mauricio Minas, Bradesco’s EVP of Technology. “Our competitors are no longer just banks; there’s start-ups, fintechs.” Bradesco’s challenge now is to talk to this new, more digitally native population in an engaging way. “And for this we rely on Watson.”

How Watson learned in 5 steps

1. Watson was trained
In Portuguese and in banking by a dedicated team with 10,000 customer questions.

2. Watson was tested
In a limited number of branches until the bank was satisfied with the responses.

3. Watson was launched
And made available to the employees of all 5,200 branches nationwide.

4. Watson got results
And as employees began to trust Watson, response times were reduced from 10 minutes to a few seconds.

5. Watson keeps learning
And improving, thanks to feedback on over 10 million interactions.

Teaching Watson the language of Brazil — and banking

Their first task was to teach Watson Portuguese. Initially there were some doubts. “It’s more than just learning the language,” said IBM Managing Director Katia Vaskys. “You also need to understand Brazil’s culture, and the regional accents, and the way each region asks a question.”

But after mastering the nuances of Portuguese, Watson was ready to be trained on the business of banking. To do this, Bradesco and IBM worked together to develop a team that taught Watson about the bank’s products and services by asking and answering questions for Watson in natural language — the same way a customer would.

“Yes, Watson can learn,” said one IBM employee, “but it needs people to teach people who are committed and patient.” Because of this team effort, Watson could understand 100% of written questions and 83% of spoken ones after just 5 months of training. After 10 months, the system was answering 96% of all questions correctly.

The Watson advantage

Now Watson is trained on 62 products and answers 283,000 questions a month with a 95% accuracy rate, with just 5% requiring calls for further assistance. In some cases, response times have been reduced from 10 minutes to just seconds. “It’s a real wow factor,” exclaimed one manager.

This helps employees have more enriching interactions with clients, because they have time to dedicate to providing the best possible customer experience. “This is when growth happens,” said the Bradesco AI Lead, Marcelo Camara. “Our current clients notice the improved service, which in turn attracts new clients, and this is what helps the bank scale.”

What’s next for Bradesco and Watson?

“We’re reaching for the stars,” said Mr. Camara. After this initial project exceeded their expectations, he said that “we’re applying artificial intelligence to new areas to bring even more efficiency and effectiveness to the bank.” And Mr. Minas agreed: “Our challenge from now on is to see beyond. What else can Watson do?”