Speakers:

• Matthew Oakeley, Chief Technology Officer, Investec Asset Management
• Geoff Galbraith, Chief Operating Officer, MAN Group
• Michael Marks, Chief Operating Officer, Legal & General Investment Management
• Wajid Khan, Chief Operating Officer, Nemesis Asset Management

Fund Operator: How important is Artificial Intelligence (AI) and Robotic Process Automation (RPA) to the fund operations community as a whole? What is likely to drive its adoption and what could limit its uptake?

Wajid Khan: AI is here in whatever form we take it.

It is very important to be able to bring data together across platforms. If you have information stored in Excel or in a PowerPoint presentation, how can these be brought together by a simple voice command, like Amazon’s Alexa or by the computer system learning our patterns?

"It is very important to be able to bring data together across platforms"

When we get into the office in the morning when we open Word, Bloomberg and our various news feeds, AI can have it ready for us.

AI will definitely play an important role in how we can improve our cost efficiencies.

Michael Marks: I start from a perspective where if I were to redraw the organisation and we started with a clean sheet of paper, what would we look like, how would we work and what would we do everyday?

Of course, we can’t start from this clean sheet of paper unless we start a new firm, but when you think like this, you see that what you do today isn’t what you did five years ago, and this won’t be what we will be doing in five years’ time.

There is so much that we can automate, we have probably just scratched the surface today. When I look at the professionals who are applying RPA in all sorts of different ways they haven’t thought yet how far into the organisation this will go.

"There is so much that we can automate, we have probably just scratched the surface today"

For most organisations this will have started in the operations space, which is where we see frequent repetitive activities that we can easily configure.

When you look at RPA tools, you need people who understand the processes, and if you look beyond the operations aspect, it is going to stretch all the way through our organisations.

The more we move away from the frequent, repetitive activity, the more we look at the AI implementations such as natural language processing and the questions that you can pop into your chatbot which pull information from a range of different sources.

Matthew Oakeley: I am quite sceptical about a lot of IT automation.

There is a great temptation for people to ‘chase unicorns’, so you have to be very careful that the end reality will deliver the benefits.

If you look at most organisations there will be a beautiful IT architecture picture somewhere that gives everyone comfort that it is all sorted. If you go to talk to anyone who is doing any real work, the relevance of that architecture picture is less clear!

When people are asked for information, they generally do the best they can with imperfect tools. When someone says, ‘Can you get me the data on X?’, the scramble starts.

"I am quite sceptical about a lot of IT automation"

You use Email and Excel; you copy, cut and paste the chart that you used from your friend last month and change some of the numbers manually; your goal is to cobble something together however you can. The big systems rarely help in the production of the data you’ve been asked for.

I suspect that this scenario rings true in more companies than would like to admit it.

The reason why Excel has been so successful is that it is a toolkit that everyone can use to do their work. We need to learn from this; if we think of automation as creating tools to empower people, it will turn out a lot better.

We need to be careful that RPA isn’t just another unicorn that we go chasing. The idea that AI will replace people is not the right mindset to start with.

We need to start by wanting to make people more effective and more efficient with these tools. Rather than RPA, there might be more use in looking at tools like chatbots, which are interesting because they offer the idea of the concierge service to help people to help themselves.

"The idea that AI will replace people is not the right mindset to start with"

If you are a mortgage processing company and you’ve got 20,000 mortgage applications coming in a month and you’ve got 100 people who take the form from one tray, process it and then put into a different tray, then there is room for automation.

However, the asset management industry doesn’t have much of this. Our automations were the Straight-Through Processing (STP) pushes in the 1990s and 2000s. We have done the big-scale operational things and we don’t have many big processes left to automate.

We need to find 1,000 little projects, which is not going to make economic sense most of the time because doing the automation will cost more than the manual process did.

Be careful not to get swallowed up by the idea of a transformational opportunity that everyone else is engaged in and that you fear you are missing out on. Be very circumspect or you will waste your money.

Geoff Galbraith: I went to India two years ago to see what the Bank of New York they were doing with robotics. All around the office they had little robotics that on the front of them had painted, “I am not Alex”. They had in fact named the robotics program Alex, but the message was that they were not looking to build a series of little robots that work around the office to replace people’s jobs.

RPA, AI and instant messaging are toolkits that if we put power to, then they drive efficiencies in the hands of users. I agree that gone are the days where there were lots of low hanging fruit to automate in the asset management world. Our STP rates are north of 99 per cent so there is not a lot of value in trying to find that last 1 per cent.

"RPA, AI and instant messaging are toolkits that drive efficiencies in the hands of users"

There may be value is in all the other pieces of work where it is more difficult to get information on what to automate – there may be difficult to spot opportunities to automate more manual tasks we are still performing outside the trade life cycle.

For example, is there any correlation of the things that we do where it is not possible for a human brain to decipher if there are patterns there or the areas that should be invested in?

Perhaps we could use AI in our Net Asset Valuations (NAV). We are playing around with AI within our middle office production cycle.

"There may be difficult to spot opportunities to automate more manual tasks we are still performing"

We have 50 people producing P&L everyday and they spend time dealing with issues around pricing and trade flows. We have solved everything that we can spot with the human brain and it will be interesting to see if there are any value patterns that we can pick up with AI.

It is not a panacea, but I do see AI as a valuable toolkit to deliver efficiencies in the future.

This excerpt is taken from a roundtable: “The take-over of machines and computers in the running of operations and the new role of humans as supervisors”.

You can read the full roundtable in the research report Fund Technology, Data and Operations, Europe 2019 here.

 

Please Sign In or Register to leave a Comment.