Historically, the term automation refers to the notion of work typically done by humans instead being conducted by machines operating within a self-governing system. In many industries, automation is not just commonplace, it’s a necessity – there’s no way that sectors such as manufacturing, automotive and medicine would be able keep up with demand without automated systems that perform tasks at a rate that would be impossible for human workforces.
But today, automation can apply just as much to the services industry, allowing businesses to hand control of business processes over to bots. And that’s largely thanks to the growth of Robotic Process Automation (RPA), something we revealed in a recent report published by The AI Journal.
Robotic Process Automation (RPA) can automate and digitize the repetitive processes typically performed by human operators, and it is proving increasingly popular in the business world.
In 2018, analysts at Gartner forecast that worldwide spending on RPA would hit $680m in 2018 and that it was on track to reach £2.4bn spend in 2022. Gartner predicted that 85 percent of large companies will have established some form of RPA by 2022. More recently, Brandessence Market Research estimated that RPA generated $2.7bn in global revenues in 2020, and is expected to achieve $18.3bn in 2027.
Applications can be as straightforward as generating an automated response to an email, through to sending out legions of bots to automate multiple jobs across a company’s ERP (enterprise resource planning), systems used across financial management, HR, supply chain management, manufacturing and distribution. For instance, RPA can match purchase orders to invoices and receipts.
RPA bots are cheap and easy to implement and do not require custom software or deep systems integrations.
By taking up RPA tools offered by practitioners – such as UiPath, Blue Prism, WorkFusion and Pegasystems – the benefits are numerous. More recently the market has seen a growth in VC $12M opensource providers such as Robocorp. Automating tasks saves time and money; allows companies to create, test and develop automation schemes within a fraction of time it would take manually.
While RPA on its own continues to enhance corporate process management, taking on monotonous tasks and freeing up people for higher-value work, it is its integration with AI that makes it an exciting prospect looking into the future. Chatbot technology, conversational AI and Natural Language Processing (NLP) are three terms that are spearheading this evolution.
RPA is markedly the deployment of rules-based bots which follow specific instructions, whereas chatbot tech is cognitive and introduces intelligence into the RPA equation.
Through the application of NLP and other AI systems, RPA systems can effectively learn through experience, allowing them to become adept at customer-facing processes, such as customer service, sales and marketing; but also proficient at managing back-office jobs.
Let’s talk chatbots and NLP
Chatbots are defined by a number of characteristics: they are applied to customer- or user-based conversations that take place by voice or text, be that phone, voice-activated interfaces, email or online chat; they are put in place to react, rather than merely automate, adapting to changes as knowledge is gleaned from data and experience; and they are intended to simulate less structured (or robotic) human conversation.
If robots had a heart, then the beating heart of Chatbots would be Natural Language Processing (NLP), the AI that governs how computer systems analyze natural language data and identify and extract meaning from contextual nuance, upon which it can then base decisions.
From a consumer point of view, NLP is already manifest in many aspects of our lives. The digital assistants that reside on our smartphones, such as Siri or Google Assistant; the auto-correction that kicks in when we tap out messages; the sifting that enables spam filters to decide what is unsolicited and unwanted email and then chuck it out; and the ability of the web to determine the intention behind our internet searches… so many elements of our digital lives are handled by NLP.
From a business perspective, the different natures of RPAs and chatbots can complement one another to a powerful degree: conversational AI can interpret customer intent, for example, and pass on data to inform more rigid RPA-driven process. RPAs can help chatbots tap into complex, data-driven requests, while chatbots can make the user experience a more natural, human experience.
For example, if an employee wanted to search for specific items within an RPA system that matches purchase order to receipts, they could use a voice-activated chatbot interface to search for information. Or let’s say a business customer wanted to book an appointment with a company but found that they cannot get through on the phone. Instead of waiting on hold, they could call or text a chatbot, which would then activate an RPA, which schedules an appointment based on the appropriate management system.
A marriage made in heaven
The business case for employing chatbot technology and marrying it with your RPA systems is a no-brainer, but firms can find their first chatbot hard to implement. Companies can use chatbots to transform user interfaces by using text-based, social media-style interfaces that allow staff to engage with ERP (enterprise resource planning) systems by simply stating their request. They can foster customer loyalty by using conversational AI to bypass call center comms and handle customer service communications in a consistent manner, 24-7 and via an array of channels or platforms (such as Google Assistant, Apple’s Siri or Facebook Messenger).
While RPAs eliminate the need for staff to do tedious administrative tasks, chatbots can talk to and understand a person using NLP. Both are powerful forms of AI in isolation. But it is the combination of the two, the ability to build conversational process automation into a business’s data-rich systems and processes that makes it such a powerful proposition.
Tom Allen, Founder, The AI Journal