Automation isn’t always about connecting APIs behind the scenes. What if there isn’t an API available or the integration would be too difficult?
One alternative is robotic process automation (RPA), where AIs observe humans using software and then learn to use the software themselves. In contact centers with legacy tooling, RPA could be one way to introduce AI to the contact center without having to replace proven yet inflexible systems.
In the sci-fi series Futurama, sentient robots live side by side with humans. Many of those robots have jobs, such as bending steel or delivering items. Often, the robots of Futurama appear to be designed to directly replace human workers on a one to one basis.
Okay, it’s a comedy but the robots of Futurama show one of two ways that robots could go. In the world of Futurama, robots fit directly into the place of humans by, for example, operating a machine that was originally designed to be used by a human. The alternative is robots that are designed in a way that rethinks the whole idea of how to get the job done. Take the robotized production line of a car factory, say.
Robotic process automation is a bit like those robots in Futurama. Rather than having to buy into entirely new software stacks, RPA lets you automate your existing software.
How RPA Works
RPA has its origins in screen scraping tools. Screen scraping is where software scripts load websites and extract content from them.
For example, in the early days of internet banking, tools began to appear that would aggregate all of your bank and card accounts in one interface. They usually worked by having software log in to your various internet banking accounts and then extract your balance information from the web pages.
Such screen scraping was clunky, at best. The developers of those bank account aggregator tools would have to manually configure their tool to know how to navigate each bank’s interface and where to find the relevant information in the HTML. If a bank made a change to its interface then the aggregators would be stumped as they had no “intelligence” to try something different. Not only that, but a lot of site owners disliked the idea of third-parties automatically extracting data from their websites and that led to captchas. If you’ve ever wondered why Google occasionally asks you to confirm you’re not a robot, well, that’s why.
Robotic process automation takes screen scraping to the next level. Rather than relying on a rigid set of rules, RPA tools observe humans using software and learn themselves how to perform tasks.
Why RPA for Customer Communication
Over the next decade, artificial intelligence will change how we think of customer communication. Already, simple bots are helping to handle customer queries.
There is a problem, though: AI tools need a way to integrate with existing systems. Depending on company size, investment in existing tooling can be significant and a migration too costly to consider. What, then, if your contact center software doesn’t allow for easy—or cost effective—integration with third-party tools?
Let’s look at an example scenario. In the next five years, we’ll start to see increasingly intelligent artificial agents directly handling customer contact. For straightforward queries, an AI-powered bot with a speech interface could handle telephone queries at a fraction of the cost of human agents. But let’s say your company’s customer management system doesn’t offer an API to integrate with the AI-powered bot. What then?
Using RPA techniques, the bot could log into your customer management system through the web interface and update the customer records itself. As far as the legacy customer management system is concerned, the bot would be just another user of its web interface. But the AI would be software sitting on a server somewhere.
What sets RPA apart from basic screen scraping is in how it learns to automate processes.
How RPA Robots Learn
Learning is at the heart of artificial intelligence. Machine learning—that is, rapidly trying many scenarios and then remembering those that result in success—is at the heart of the natural language understanding and speech recognition used by chatbots and virtual assistants such as Alexa.
RPA relies on humans first showing the software how to do something. The human completes the required tasks multiple times while the RPA tool observes what is happening. It then builds a picture of the workflow. Human operators can then modify the workflow or even seed it at the beginning with the optimal workflow. But the important thing is that the RPA tool never stops learning.
If, for example, the “Submit” button on a form changes to a “Go” button, then the RPA tool should have enough data and independence to give that a try. As time goes on, the RPA tool can learn more and more tasks, either by observing humans or by making its own deductions.
First Steps to RPA
This might sound like a recipe for replacing humans; actually, RPA today is more about making better use of human agents.
Just like all robots, RPA excels at helping humans to avoid repetitive work. So, in the near future we’re more likely to see RPA tools that make case notes for a human agent, a smart digital assistant, rather than the RPA tool handling the entirety of the interaction and thus replacing the human agent.
As we automate more and more customer communication, RPA will help to fill in those gaps where other forms of integration aren’t possible. It’s a great example of pragmatism in how we approach changes in business. In principle, automation offers enormous opportunities to save on costs and deliver better customer experiences. Practically, though, it’s neither possible nor desirable to wipe away legacy systems and start with a completely fresh slate.
For customer communication, RPA offers an opportunity to make those legacy systems part of your omnichannel conversation with customers. Perhaps you want to create a WhatsApp chatbot that enables customers to access data that’s only available through a legacy system. Using Nexmo’s Messages API and an RPA agent, you can make it happen without the need for expensive re-engineering.
Robotic process automation is just one of the ways in which you can automate your customer communication. Take a look at my other posts on Automated Customer Surveys and Bots, Assistants, and AI for more ideas.