Last month, the pharma company Takeda began recruiting patients for a clinical trial of a promising Covid-19 treatment involving antibodies drawn from the blood of recovered patients. It normally takes several weeks to collect people’s information, determine who may be suitable for the trial, and get the paperwork in order.
Takeda started testing RPA, several months before the pandemic, with software from UiPath. “We’d been proving that there was value to it,” says Kyle Cousin, head of Takeda’s digital service line and the person in charge of the effort. “Then around Covid we said okay, we can accelerate drug discovery and get patients through the cycle faster.”
Inspired by the success, Takeda is now stepping up its use of RPA with a plan to train thousands of staff to build and use software bots for themselves. It recently ran a successful pilot with 22 employees. It estimates that the effort could automate 4.6 million hours of office work per year—the equivalent of roughly 2,000 full-time workers. But Takeda doesn’t see the technology displacing anyone. Cousin says the goal is to boost productivity, and hiring has increased as the software bots have been rolled out.
For all the hype around artificial intelligence and machine learning, the quickest and easiest way for companies to automate office work is through simple and decidedly unintelligent software automation. Takeda’s approach provides a way for machines to take over routine and repetitive tasks without investing in a big software project or worrying about legacy systems. It’s hardly elegant or robust, but as long as you can point and click, you can automate.
The trend serves as a reality check for the AI industry. After years of hype around the commercial potential of bleeding-edge machine learning research, COVID has caused some hard-hit companies to rein in blue-sky investments. Last month, Uber said it would wind down its AI research lab as part of broader cost-cutting measures. At the same time, as companies face pressure to do more with less (including fewer employees), interest in simple-yet-effective automation may grow.