Virtusa
The back office is a robot’s world, according to Senthil Radhakrishnan
of Virtusa. And artificial intelligence could mean even more opportunity
How is Virtusa using robotics for lifecycle management?
We focus primarily on client onboarding, know-your-client (KYC) regulation and customer relationship management. In retail banking, a lot of that is already automated, but for institutional onboarding, especially considering the amount of new regulations coming in to effect, there is some special attention required.
Previously, regulators have been quite relaxed with this kind of thing, but in the last few years they have started to pay much closer attention and there have been a few very large fines issued. At the same time, however, banks are realising that there will be a business advantage to improving these processes.
Virtusa used to help banks come up with onboarding workflow solutions and introduce automation to those systems. Many of our clients had huge operations teams, and staff members would have to physically phone their clients to collect the data and then populate the systems. We introduced data vendors, so banks could pre-populate their data systems, only contacting their clients to collect differential information. That way, they gained good, clean data, and removed the risk of operators making mistakes.
Now our clients have matured. They have KYC policies in place, they have working systems and they are meeting regulatory demands. Now, what we can help them with is reducing their costs.
The thing about any client onboarding is that it’s not just a one-time thing. Regulations require banks to keep revisiting and verifying client information—every three years for low-risk clients and every year for high-risk clients. That involves a lot of work and, at the moment, a lot of money. But with robotics and specialist software, banks can potentially reduce the cost of processes and the headcount of their operations teams.
Another pain point that we hear from clients is that it is difficult to get a unified view of the clients coming on board. There are many different systems, and different onboarding dealers, or a particular team that will manage all new institutional clients. They’re the only ones who have a view of that client and that client’s data, and although they might update spreadsheets and liaise with that client, the front office won’t have that information, and so they might be trying to source information that is already available. We are trying to introduce ways that anyone authorised can access that information instantaneously.
To what extent will robots replace manual processing?
When there is a process, such as KYC verification that has to be repeated every year, or every three years, that is multi-step and repeated multiple times. Staff members will run web searches and look through regulatory websites and data vendor sites to see if a client’s status has changed—those processes could be completely automated. A robot could be trained to do the whole thing.
Robotics can solve a lot of problems. Today, skilled people are doing tasks they shouldn’t be, and a lot of their processes can be broken down in to smaller processes and given to a robot, leaving the team free to complete the niche, analysis-oriented tasks. Banks can still reduce headcount, but they will also utilise their people better, as qualified people won’t be doing mundane tasks.
It’s important to point out that we’re not trying to say robotics will fix inefficient processes. Giving the same process, as it is, to a robot will still make it faster, and still makes sense.
In fact, in a lot of cases, fixing a process itself means huge investment, which means banks will hesitate. They know that if they spend the money they will be able to improve the process, but they don’t want to spend, so instead they continue working with it the way it is and end up with a huge operations workforce doing things in an inefficient way. Robots can be trained in a couple of weeks, so the banks start to see results quickly. The cost is an annual licence, so there is no big capital expenditure and no uncertainty about when, or if, they will see a return.
Where does artificial intelligence come in to it?
The majority of banks are still exploring in this area—they have heard about artificial intelligence and they want to know what they can do with it. It is a buzzword, but it hasn’t got much further than that yet.
There are still some questions about how quickly and efficiently a robot can be trained, and whether it could just observe a human at work and train itself. If it is given data in a format that is not very defined, could it still glean information from it? These are elements of machine learning that could be used to make data processing much more efficient.
What is interesting is that artificial intelligence is non-deterministic. Robotics is deterministic, so you can give a robot a task and it will do it. That is a no-brainer. But with artificial intelligence, you have to think quite carefully about where you want to use it, and how. You can’t let it work independently—it will have to come back to a human for review, but it could make that person much more efficient. That is what we are focusing on.
If we take KYC compliance as an example, the regulatory mandate is that if a news article comes up that mentions a particular client, a bank has to act on it, or at least be aware of it. There can be a lot of data coming in from different sources, and that can be too much for a human to deal with. Artificial intelligence can be used to gather that data and make some sense of it. The machine can flag anything unusual, where a human might have missed it, and the human can act on it.
Will data collected by machines be more accurate?
Bad client data is a major issue that banks are facing today. A bank might have a large corporate client that has one account in the corporate bank, one in the investment bank, and another account for a branch of the same client in a different country. All of those accounts might be completely separate, with no single view of that client available, even though it’s one corporate organisation.
Banks are starting to sort out their client data to de-duplicate. So far, their approach to this has been to search for particular terms for a client and to combine any documents that contain those terms.
With machine learning, they can point the software towards the data to look for similar clients and data sources, and it will start throwing out results. A human could look at those results and pick out what is relevant, and after a while the machine will stop including certain things it finds—the machine will start to learn what is correct and what is not. The more sophisticated software will also identify similar terms or misspellings, and the technology is only getting smarter.
Big banks are looking in to this themselves. Whereas they would usually wait for an established company to bring out an offering and then look in to it, they’re starting to work with start-up technology companies, integrating the technology in to their systems directly.
Is artificial intelligence likley to change the face of financial services?
There are computers out there that can play chess, predicting moves on the board. Some firms are looking in to this in terms of wealth management—they’re looking at portfolio data and investment opportunities, exploring the idea that software could predict trends, tell managers when they should rearrange their portfolios, and how to get higher returns.
It could also potentially identify impending risks to portfolios because of various environmental factors. These predictions are getting more and more accurate, and again, the machine would only get smarter as it goes along.
With technologies such as blockchain becoming more important as well, cost structures are going to change and whole entities—whole companies—may no longer be required.
Banks are investing a lot in technological research at the moment, and artificial intelligence has not fully matured yet, but banks may be about to find themselves in a purely supervisory position, with very little role to play in their own back offices.
