Robotic Process Automation - the quick fix for process efficiency

  • Rob Yeates, Software Engineer at Gresham Technologies

  • 18.02.2018 06:15 pm
  • undisclosed

As enterprises come under increasing pressure to digitize operations, Robotic Process Automation (RPA) is gaining momentum.

The reasons, according to Forrester, are three-fold:

  1. It boosts productivity with minimal process change
  2. It offers an easy-to-calculate ROI
  3. It’s a fresh alternative to the big spend of a typical BPM programme.

We would argue a fourth benefit:

  1. Increased data security and a clear audit trail.

However, during a recent insurance breakfast hosted by Gresham, the subject of RPA was raised and it became clear that everyone attending had a slightly different understanding of the topic.

So what exactly is RPA - which problems can it solve and which problems can it introduce?


Business Process Management is related to RPA but has significantly different goals. Where RPA is often a surface level fix to automate existing workflows, BPM is a wholesale re-imagining of individual business processes to improve efficiency.

Business Process Reengineering, meanwhile, involves even deeper change; a recreation of processes from scratch, that can often impact across a business’s entire operations.

RPA is a surface-level fix where BPM and Reengineering look to solve the underlying problem itself.

Why is RPA needed?

In financial transaction processing, a number of functions still rely on manual interactions. Let’s take an insurance broker and the insurer statement as an example.

  • Policy, Account and transaction level data all need to be reconciled.
  • The matched data then needs to be posted downstream to a Policy Administration System that consumes it.
  • The Insurer also needs to be notified of matched transactions authorised for settlement, often via an online portal

Reliance on manual system updates introduces significant risk into the process (not to mention additional cost). A single mis-keyed digit could affect the integrity of the entire transaction. 

A similar risk exists for our capital markets customers where data between traders and prime brokers is reconciled. There is inherent risk if matched transaction data is keyed into the stand-alone UI of a legacy position management system.

RPA can automate this part of the manual process; consuming from a data source such as Excel, logging in to the position management software, navigating to the correct screen and keying in the data, taking specific actions and reacting to the alerts and prompts of the target system.

There’s no manual requirement, data quality is guaranteed and there’s an audit trail into every action taken. It’s also quick - to reimagine this process with BPM would be cost-prohibitive and difficult to justify; with RPA it could be live in days.

How does RPA work?

  • Optical Character Recognition (OCR) is used to convert images to text - so for RPA you can convert a PDF into usable data, to be automatically entered into a UI.

Taking our insurance example again, an insurer could send a statement by post, or as a PDF attachment within an email. Instead of manually keying in the data, OCR identifies and extracts the relevant pieces of text, for example the account details and identifiers of the transactions together with the amounts.

OCR is used widely across almost all of Gresham’s Clareti applications to extract critical data using this method. 

  • UI automation can take the data produced through OCR to automate character entry and button clicks, removing the need for manual input.

It is not a new area of development, but enables the robotic exercise of any given UI or front end. Gresham uses a number of UI tools in its software - for example Selenium, to test web front ends. This form of RPA carries out end-to-end testing of the core platform, simulating thousands of discrete user interactions in seconds.

  • User configured workflows; comprising a number of low level commands can be created using a flow diagram. 

Returning to our capital markets example, the workflow might be:

  1. Is email present
  2. Get email attachment
  3. Read values from PDF
  4. Login to Position management software
  5. Was the login successful? If no, fail and alert
  6. Navigate to trade entry screen
  7. Enter trade data
  8. Click ‘save’
  9. Confirm data is present in separate screen. If not, fail and alert
  10. Repeat
  11. Logout
  12. Notify on success.
  • AI is coming, but according to Forrester is not yet present in most RPA tooling. The ability to understand an issue and make a decision will increase the scope of RPA functionality, though it remains to be seen what this will look like. There would need to be a focus on the auditing of automated decisions and actions - particularly when RPA robots will be chained together.

Are there any issues with RPA?

RPA has the potential to make an automated process brittle. Without the knowledge of what acc_fr_ex_brk_gi actually means, it becomes more challenging to rebuild a process.

Equally, by dealing with the symptoms of a poor process - applying a sticking plaster instead of dealing with the root cause, there could be compromise down the line. The decision to use RPA to fix today’s problem, instead of investing in rebuilding and improving it, isn’t straightforward.

Regardless, with the increased abilities that AI will introduce, RPA will doubtless continue its stampede throughout the financial world. For those looking to streamline processes - an evolution, rather than the revolution of BPM or BPR, the potential use cases are innumerous.


Gresham and RPA

The value of RPA is certainly one that Gresham subscribes to (and has done since day one - so we’re slightly surprised by excitement over its ‘new’ potential).

Workflows within Clareti Transaction Control (CTC) already offer a number of valuable automation steps - and it uses machine learning to take RPA one step further. For example, if a user completes a manual workflow action and the same scenario arises again, CTC will suggest to the user that the workflow could be automated through business logic.

CTC already achieves industry best match rates and we are further extending our lead by incorporating machine learning techniques which enable us to understand how data was matched and provide pre-trained models to drive even higher match rates.

Integration with RPA tools will increase in importance and the open nature of the Clareti platform means integration in straightforward. CTC has been connected to Blue Prism at customer sites and these skills now exist within our professional services team.

Other Blogs