Amelia
Amelia, the friendly cognitive virtual agent

Accenture has chosen an artificial intelligence-powered virtual agent called Amelia, to push cognitive machine learning towards businesses like banks and insurance companies.

Amelia, developed by IT automation provider IPsoft, is a self-learning cognitive agent you can actually talk with in natural language. In other words, she can hold a back and forth conversation and, as well as answering questions, she can manage processes for you. For instance, in banking, she could open up an account for you.

Edwin van Bommel, chief cognitive officer at IPsoft told IBTimes UK: "Amelia will answer your questions in natural language. It will all be very conversational, as opposed to AI-based services like Siri where each utterance is recognised as a new separate question."

On top of natural conversation capability is Amelia's learning ability. To teach Amelia about how to answer calls in a call centre, call transcripts are uploaded to the platform.

Amelia can provide a service to consumers or to business clients. She also has internal applications, such as working as an IT help desk. Another internal service is providing staff with HR services, answering questions about pensions and so on. Thirdly, she can answer banking employees' questions about some dos and don'ts regarding regulation.

In terms of performing automated processes, Amelia can build on analytics, or gather expertise, said Van Bommel. "It's quite easy to link Amelia to some of these systems or put some of these knowledge elements into Amelia, but it depends a bit on what type of analytics."

Accenture believes the AI and cognitive systems software market will grow from £3.1 billion in 2014 to £6.4 billion in 2019, with others citing these systems as a catalyst to have a £3.5 trillion – £4.9 trillion potential economic impact by 2025.

In terms of how services like banking might evolve with this sort of technology, Van Bommel said: "Let's say in your new bank you open up the app and the app is actually empty and there is only a chat window. With this chat window you can do everything with your bank − ask questions, ask for a service.

"Let's say you lost your wallet and your cards need to be blocked and you need to get new cards. You don't need to look in a menu, you don't need to look for phone number.

"You just get the app and say your wallet got stolen can you help me? The cognitive agent will take you through all the processes and probably be very empathetic. Amelia can also respond to emotions.

"Although you know it's a machine, you will appreciate that it's a machine that is actually helping you. We don't pretend to be a human; but it's a machine that can help you in a very fast and comfortable and also personal way.

Van Bommel said that interactive voice response (IVR) scores very low when it comes to helping people with problems.

"On average questions were taking 18 minutes. Amelia took it down to 4.5 minutes. We found that if clients have 24/7 access and are being supported very fast by a machine that is not completely stupid, they actually appreciate it much more."

According to a statement, enterprises are already engaged in integrating Amelia into their businesses. For example:

  • A global oil and gas company has trained Amelia to help provide a prompt and more efficient way of answering invoicing queries from its suppliers.
  • A leading European bank has completed a rapid deployment of Amelia within the IT function, successfully equipping the cognitive agent to take action a mere 45 days after project start.
  • A large US-based media services organisation taught Amelia how to support first line agents in order to raise the bar for customer service.
  • A prominent global bank intent on seizing first-mover advantage in its digital strategy has successfully tested Amelia's ability to support its network of mortgage brokers by providing guidance on policy details.