If you haven’t heard of Robotic Process Automation (RPA) directly, you likely have heard of some of its major players:
And, of course, Xerox and Microsoft. A few years ago, RPA vendors and industry analysts were hyping up RPA to be the next big technology: “It’s not hard to see the opportunities [for RPA] everywhere once you get engaged,” stated the former vice president of Gartner.
Nowadays, more people are investigating how RPA and AI can ‘play nicely’ in internal company workflows. RPA offers a compelling advantage – the ability to construct automatic workflows. Artificial Intelligence (AI) also enhances RPA with an extra ability – continuous learning.
RPA provides a robust foundation for the increasing capabilities of AI. However, RPA and AI products must combine these technologies carefully, ensuring that AI isn’t a gimmick or an add-on (an example of the increasing phenomenon known as AI washing). Let’s examine RPA’s increasing capabilities over time.
In summary, increasingly evolving RPA and AI solutions have combined with existing financial systems to create sophisticated, specialised tools for financial services. However, AI has been cultivated by academic and industrial experts for a long time and has received more buzz since ChatGPT’s release in 2022. Are RPA and AI just two (extremely helpful) technologies, or are they actually rivals?
The simple answer – they are allies. We’ve even written an article describing the top seven benefits of RPA and AI partnerships. In finance circles, RPA and AI have collaborated on various high-profile projects, such as HSBC’s customer service query (BOLT) and Hiscox’s lead underwriting model, powered by Google Cloud and UiPath.
However, RPA and AI’s companionable status is conditional on not mistaking the technologies as synonymous. RPA is older, more established – and more limited. Though RPA can automate repetitive tasks, it lacks the flexibility to make decisions and solve problems. In contrast, AI’s capabilities are constantly evolving. It’s becoming faster, more adaptable and cheaper to operate.
What do RPA/AI fusions actually look like? Let’s explore a few examples of tools that incorporate RPA and AI technical elements.
Many document workflows classify the document, extract the data and then action it. The process is smooth, with minimal manual touchpoints.
RPA can move data along in an automated process. For example, when we worked with Enterprise RPA, their tech stack captured remittances and invoices from customer emails.
AI is particularly adept at data extraction, as it can understand the context of raw data. When AI (occasionally) makes a mistake, if manually corrected, it will learn from the flagged data to never replicate the error.
Predictive analytics refers to tools that model or forecast performance, such as financial, operational, legal and marketing tools.
RPA must ensure that predictive models are fed with a flow of high-quality, topical data to produce the most accurate predictions. How they do so is more nebulous, as it depends on the RPA model. Some will automatically enable text recognition, clean the data or complete feature engineering.
AI is excellent at selecting and executing advanced statistical techniques for analysing data. Machine learning is particularly useful for predictive analytics, leveraging techniques like regression models and decision-making to identify the underlying patterns in data.
Chatbots became trendy and widely available in the 2010s as firms realised that they could be significant revenue-boosting tools. RPA and AI together can craft more sophisticated chatbots than following simple mechanical workflows.
RPA integrates virtual chatbots with existing company systems and AI and machine learning technologies. RPA is particularly useful for companies whose legacy systems do not have the capability to connect via APIs.
A subsection of AI known as Natural Language Processing (NLP) can be deployed to analyse language to detect sentiment and intent. Users will be pointed towards the information they’re looking for quickly. Plus, AI will harvest the anonymous user data to improve their experience continually (e.g. by tracking chatbot usage patterns and improving the quality of the bot’s responses).
For the past four years, we've collaborated with Enterprise RPA to process remittances and invoices. Our technology extracts the relevant data, which is then fed into their workflow to automate payments.
Enterprise RPA's sophisticated AI and RPA-powered tech stack has enabled them to continue expanding its client base, providing sophisticated automation solutions for esteemed customers such as Hellman’s Worldwide Logistics, Southern Housing and Aspire Housing.
“The next generation of automation must do more than just sit on top of legacy systems,” Steve Morgan, Global Banking Industry Lead at Pegasystems, explained to FinTech Magazine.
Why is this an intriguing quote?
It’s true that financial firms could adopt AI and RPA automation gimmicks (such as chatbots or AI-powered customer relationship management software) without fundamentally changing their software. And it implies that the adopters – and developers – are equally responsible for the future of automation.
Integrating RPA and AI algorithms meaningfully (rather than just placing them ‘on top of’ existing automation) means considering how business processes can be redesigned from the bottom up. It’s a drastic change, especially for large enterprises with delicate cross-silo relationships. But using these tools can achieve more than shaving off the occasional few minutes by using ChatGPT directly.
Unlike technologies with more complex relationships (like OCR and AI), RPA and AI can be combined to create versatile automation solutions. As our partner relationships have demonstrated, AI and RPA can work harmoniously. With some testing and tweaking, AI and RPA’s results are tangible and powerful.
Interested in hearing the latest insights from AI? Follow us on LinkedIn.
