01 Apr, 2019· read

Complex business problems need multi-skilled software robots


In our last blog we talked about the evolution of automation and how adding intelligent skills to RPA could be a key way to extend the scale and benefits provided. We also provided an overview of the intelligent skills our Digital Experts are equipped with: problem solving, machine learning, collaboration and optimisation.

While these skills deliver value in their own right, it is the dynamic combination of these skills which is critical to solving complex business problems. As humans, when faced with a complex task, we observe and assess the problem, diagnose cause and perhaps rank the importance. We then apply our knowledge to identify the best action, and if we’re not sure, we ask someone! Together these skills represent a broad scope of intelligence. To truly create intelligent automation solutions, we must replicate (and enhance) this human capability.

For example, let’s consider fraud detection in the Insurance industry. Essentially this is about continually improving the capability to identify and act upon fraudulent activity. To do so requires an approach that can capture information, understand propensity and act, all on the basis of known or learned intelligence.

A Digital Expert can be trained in this combination of skills to automate fraud detection. For example:

Problem Solving – the Digital Expert could monitor each claim and assess it against company policies to identify when it is likely to be fraudulent or contravenes with company policies. For example – a company’s policy might be to question any claim that was made within 5 days of the insurance policy being taken out.

Machine Learning – the Digital Expert could analyse all the previous cases of fraud and identify the patterns that would indicate fraud and apply them to each claim. For example – machine learning might identify that when there are multiple injury claims for a single claim it is likely to be fraud. Machine learning delivers an accurate probability score for fraud which can drive a corporate action; for example, a score over 25% may always be referred to the fraud team.

Collaboration – the Digital Expert could advise an assessor or hold a conversation directly with the claimant to gain further information relating to the claim. This would enable capture of information not provided in the claim form – did they sign themselves off work? AI services such as sentiment analysis could help further in detecting how genuine the claimant sounded.

Together this combination of skills allowed the Digital Expert to build an understanding of the likelihood that a claim was fraudulent – illustrated as a confidence or probability score. The insurance company can then take the most appropriate action as a result. This allows the human co-workers to focus on the really valuable activity of investigating the likely cases of fraud rather than processing that transactional activity of reviewing each claim in turn. The scale of the automation solution was extended and enhanced - increasing productivity and efficiency further, but also job satisfaction of the teams, and even mitigating the risk that fraudulent claims are missed due to the volume of claims received.

Get in touch

For more insight and information relating to intelligent automation please get in touch

Or please come and speak to us at Blue Prism World 

Bpw Partner Twitter 1024X512

Are you ready to
start a conversation?

Get in touch