Ryan Moore, Head of Data and Analytics, explores how using AI-based technologies can help you to get up and running with RPA much faster.

As COVID-19 creates a ‘new normal’ in our lives, companies are increasingly faced with staff shortages in contact centres and IT support systems, losing offshore teams overnight due to the impact of lockdown. Some companies are actively using this as an opportunity to revisit their automation projects which were previously put aside as low priority or stalled due to a lack of funding.

A simplified kind of automation has been around for ages with programmes such as Windows Task Scheduler and Automator for Macs. One automation technology in the spotlight now is RPA, or Robotic Process Automation. RPA is bringing with it the next big wave of fast, cheap and scalable automation that will transform industries across the business world.

What is RPA exactly?

RPA technology allows users to configure one or more scripts (also known as ‘bots’) so they emulate tasks that a human would normally perform manually – for instance, inputting data, triggering responses or executing transactions. Increasing numbers of companies, like UiPath and Blue Prism, are bringing RPA to businesses. In fact, according to Gartner, the RPA market grew more than 63% last year, making it their fastest growing enterprise software category.

So, you want to automate a process using RPA, but where do you start?

In the current climate a lot of businesses just don’t have the luxury of setting up elaborate programs of work to roll out automation projects. Here, I’ve capture an overview of how to identify and trial RPA on a single business process:

  1. Identify which candidate processes to automate. You can start with a list of key processes and then score them based on cost impact, volume and human effort required to complete. At this stage you need to determine whether you have a transactional, rules-based task or not. RPA works best where structured data has clear, pre-defined rules and parameters - for instance, creating invoices. But businesses don’t always have this data in a structured format. This is where technology like InsightMaker, our information discovery platform, comes in to help you overcome that.
  2. Make sure you get management and stakeholder buy-in. Present back the candidate processes along with their scoring and link it back to Return on Investment (ROI). Think about how FTEs (Full Time Employees), released from this manual process, can be redeployed to other teams and processes.
  3. Choose your RPA partner and develop your solution. Talk to subject matter experts to create a process map, if one doesn’t already exist. This will define how your RPA bots will automate the workflow. Spend time in your contact or operations centre where the bot will be deployed to see how it will actually work in the real world.
  4. Decide what kind of bot you need. Automating the workflow will vary depending on the tool or platform you go with but, in general, RPA platforms offer you interfaces to create bots, run them and monitor them. As part of the creation you will need to determine whether you need an attended bot, an unattended bot or a combination of the two.
  5. Test your solution. Pay close attention to non-functional requirements and multiple business scenarios. Follow this up with a few short sprints to get a finished RPA bot. Most RPA deployments take less than two months to complete, but this timeframe obviously varies depending on the business case.

Where RPA meets AI

One of the main blockers to implementing an RPA solution is a lack of standardised processes and structured data. Not all processes are RPA friendly and this is often down to the unstructured nature of data that sits behind these processes. Very often, RPA projects are expected to curate data from both semi-structured and unstructured sources (such as documents, emails, webforms and chat transcription), turning it into structured requests that RPA bots can work with. At Aiimi, we’ve built a solution for this problem.

Using InsightMaker's AI capabilities to turn unstructured content into structured data for RPA bots

Our software InsightMaker comes with built-in machine learning models that help turn unstructured content, emails and chat transcriptions into structured requests that RPA bots can work with. Under the hood, we use a combination of OCR (Optical Character Recognition) image analysis, natural language processing, intent classification, and other machine learning techniques to make this happen. InsightMaker also helps classify and prioritise the requests that are most urgent, such as your most important customer queries, so the bots can automate your tasks most efficiently.

Dealing with the challenge of unstructured content is one of the best ways to get your RPA solution up and running, fast, to boost your team’s productivity and better serve your customers.

Find out more about how InsightMaker works and discover our three solutions to make your organisation’s transition to a remote working environment easier.

Header Photo by Franck V. on Unsplash