AI and machine learning (ML) can undoubtedly help when it comes to interpreting and classifying data, sending it where it needs to go, and identifying exceptions or gaps. It can also decide on the next best actions to solve such issues.
But both rely on learning from experience. They start with an algorithm that tells them the basics and work from there. The more data and situations they experience, the more accurately they operate. It is a process of continual improvement.
Indeed, RPA bots, or digital workers, can act effectively as mentors. Initially, by programming a bot to interact with systems by mimicking the keystrokes and movements that a human performs on a keyboard, digital workers automate processes by deciding what to do with something and then doing it. They work to automate manual, repetitive, high-volume, and rules-based processes involving data in a structured format. RPA can also automate access to legacy systems that lack a modern API.
The AI and ML learn from the pre-programmed RPA until they get to the stage where they know enough and have learned enough from it to go it alone. At that point, the AI and ML can start telling the RPA what to do!
The RPA/ AI combination is massively helpful in instances where a lot of disparate data needs to be taken in and sent to the right place. Indeed, managing document filing and storage can take up a lot of time and money if not properly managed. Even if filing is done correctly, doing it manually is slow, and further, finding and replacing missing files wastes even more time.
Data management is all the more pressing given the huge amount of documentation that firms need to store. There’s also a regulatory angle in that firms need to demonstrate to the regulator that they are storing documents correctly and show a proven document management trail at any point in time.
This can involve masses of data and documentation, all of which need their source and journey tacking and then storing appropriately. The task can quickly become unwieldy if not tackled properly right from the beginning.
There are plenty of instances where unwieldy data and processes need fixing to go in wealth management, at the front end from onboarding, KYC, due diligence, and the like, as well as back-end systems such as custody and reporting and compliance.
Onboarding is probably the obvious example to use as a case in point and is certainly one of the most commonly cited pain points when it comes to data and processes. Winning a new client is just the start of the relationship and, sadly, can also be the end of it if the onboarding process is not up to scratch. It should be an opportunity to showcase what the client can look forward to, but instead, it is often an exercise that is drawn out, where requests for data are repeated and where there are frustrations on both sides. Drop-off rates are high!
Onboarding is tricky because it touches on many areas, including KYC, AML, risk, personalisation, lifecycle management, and broader contextual information. The issue is that data has been previously in distinct silos in different formats. Using RPA as a framework for AI/ML, some wealth managers are improving the process exponentially.
RPA can deal with all this. It can identify what data is or refers to and then send it to where it needs to go. It can see what is missing, ask people or systems for data or information, fill those data gaps, create profiles, and arm the adviser with the next steps to complete onboarding-type information.
The task is to combine things from different areas and digest them to create meaningful process management and output. This reduces cost, promotes growth, and enhances the client experience.
Down the line, this means that the wealth manager has a clean and single database that can be used to pre-prepopulate forms in other areas, say using risk and due diligence assessments for Consumer Duty purposes.
In the bargain, the smooth and robust data collection and validation process can also be extended to other functions within the firm, notably client lifecycle and personalisation.
Once the RPA has done enough work to inform a proper machine learning algorithm, then the AI/ ML can take over and add to what the RPA has been told to do and not do. It can start to make some decisions about where things belong and automatically file things. The RPA still has its place in carrying out the tasks mandated, but instead of them being mandated by humans, machine learning takes over that part- giving the RPA the instructions, and the RPA then carries them out.
Obviously, there still needs to be some human oversight; the AI/ML should know what it does not know, and care is needed regarding false positives and negatives, and other biases.
However, the combination of RPA and machine learning has a significant role in improving client experience and internal efficiency when it comes to onboarding. A streamlined and automated solution within the onboarding process is a value add not only in terms of the actual asset transfer when a client onboards but also in terms of ensuring the ongoing client experience is enriched. This makes for client stickiness and a greater share of wallet, which are, after all, the end goals!
