Discovery and Exploration

From RoboChat to RoboBrain: UBank uses Watson to enhance customer satisfaction

Share this post:

At UBank, we think of ourselves as “fintech with a banking license” — which means that we’re an online bank that uses technology to make simpler, better and smarter banking products for our customers. When we were exploring how AI could help us create new opportunities, the more we learned about Watson, the more we realized the potential to help us solve existing challenges. AI gave us the opportunity to tap into the data we had at our fingertips and make it even more powerful for the benefit of our customers while also improving the productivity of our team.

We started our AI journey with Watson Assistant, using it to create RoboChat, Australia’s first virtual assistant designed to help with the home loan application process. With the power of Watson Assistant, RoboChat responds to customer queries on 40 core home loan topics, along with hundreds more associated questions. Initially trained on data collected from customer questions submitted through UBank’s LiveChat experience, RoboChat continues to learn as more customers engage with it.

Our first AI solution delivered strong results:

  • We launched after only 6 weeks of development work
  • As of mid 2018, RoboChat has now answered 31,000+ customer questions, and 4 out of 5 people are happy to use RoboChat
  • When customers engage with RoboChat, we see a 15% increase in conversion through the online application form
  • RoboChat now answers more than 80% of questions correctly the first time

With that kind of success, we wanted to translate it into a new AI opportunity for our own employees–one we call RoboBrain.

RoboBrain: an AI-powered information sourcing tool to increase employee productivity

With RoboBrain, we took an inside-out approach by focusing on how to simplify our systems for our UBank team members. The solution was developed when we combined all the information available in our various knowledge sources to create a one-stop, one-screen solution to help our advisors find answers to customer queries more quickly and more accurately. RoboBrain uses Watson Assistant, Watson Discovery and Watson Knowledge Studio, plus a custom interface for our advisors. The solution combines natural language chat with intelligent search, natural language processing and advanced text analytics to continuously learn with each interaction.

When advisors ask questions, RoboBrain searches through our knowledge base to surface “snippets” that best match the query -similar to traditional search, but one that contextually understands UBank’s products, features, processes and teams rather than relying on keywords. Advisors can rate the usefulness of the snippets RoboBrain presents, which in turn continues trainingWatson Discovery to improve relevance. Advisors can also choose to directly access the source documents. With RoboBrain advisors ask questions and get results in near real-time, dramatically saving time and effort as advisors no longer have to manually search across multiple systems and documents to find what they need.

Early days

RoboBrain launched in March this year and has been greeted with enthusiasm by our team. While it’s still a new solution for us, overall, we’re looking for improved performance in three areas: Employee satisfaction based on employee net promoter score (eNPS), reduced average call handling time, and reduced complaints and compliance issues.

The more we integrate Watson AI solutions into UBank work flows and processes, the more we’re seeing how we can solve problems we previously didn’t think were solvable. We’re eager to see what the next chapter of UBank’s Watson AI story holds.

See how teams are using Watson to increase productivity

Head of Digital and Technology, UBank

More Discovery and Exploration stories
December 10, 2018

How to future-proof your business with AI

By 2020, according to Gartner, the amount of enterprise data created will increase by 800%. As companies accumulate larger amounts of data in different formats and locations, the complexity of understanding the totality, while still being able to derive useful timely insights from it, demands a robust AI solution.

Continue reading

November 29, 2018

Audioburst pioneers personalized on-demand audio with the power of Watson NLU  

Audioburst listens to source feeds like radio stations and podcasts, and segments and indexes millions of minutes of audio in real time. With the help of Watson Natural Language Understanding, Audioburst has created the world’s largest library of audio searchable by content, context or theme.

Continue reading

November 21, 2018

Empowering agents with Watson, no matter where your data lives

With the integration of Watson Discovery and IBM Cloud Private, organizations have flexibility in AI deployment and bring Watson to where their data resides. WADex for ICP provides an ingestion engine that includes document conversion capabilities to consume files across different file types.

Continue reading