Generative AI is reinventing robotic process automation

The addition of ChatGPT-like generative AI (genAI) technologies is helping many companies automate processes that would have been impossible just a year ago, according to an automation expert who called genAI “a massive shift, not just in what we can do, but in the ease of putting together automations". 

One very common robotic process automation (RPA) process – parsing an invoice or other document for data and extracting that into an accounting system – has become dramatically easier as genAI tools replace conventional optical character recognition (OCR) based scanning, explains Nick Beaugard, chief software architect with RPA vendor TribeTech, whose World of Workflow platform enables a broad range of RPA processes.

“We’re now in a position where we can just take the raw text of that document, throw it at a large language model [LLM], and that LLM will actually be able to process that and turn it into real, actionable data,” he said.

“I spend a whole bunch of my life now going and talking with people, finding something that was ridiculously impossible a year ago, and – in a couple of hours – having it fully automated and working.”

“The introduction of LLMs is accelerating ROI for these tools and giving not just a better return on investment, but a far faster time to value.”

The pursuit of this ROI is driving businesses to embrace ROI at speed, with Gartner predicting spending on the technology will grow by 20 percent to reach $5 billion this year alone.

But where should businesses start to get the biggest wins from RPA? Processes that involve copying data between systems, or “when you have to do a lot of similar things at the same time,” are a good place to start, Beaugard said.

“RPA makes you more efficient,” he explained. “It increases quality because computers make less mistakes. And, so, getting humans out of the loop of doing these sorts of things improves the outcomes for everyone.”