Last week, Salesforce unveiled Agentforce, a suite of A.I. agents that its business clients could deploy to automate multi-step tasks across their workflows without human intervention. The software giant claims its agents can “analyze data, make decisions and take action” to complete tasks, such as answering customer service questions and qualifying sales leads all on their own, which could lead to significant cost savings and productivity gains. Salesforce is the first major tech company that has officially released A.I. agents since the generative A.I. craze began in late 2022. Unlike A.I. assistants and chatbots, A.I. agents can autonomously make decisions and perform complex tasks like problem-solving by understanding the context in real-time. However, the use of the so-called “digital workers” that automate manual tasks is hardly new in the tech industry. SS&C Technologies, a global provider of software and services to finance and health care clients, has been deploying these digital agents since last year and has seen huge savings.
The A.I. agents market is currently valued at $5 billion and is expected to grow to $47 billion by the end of the decade, according to the research firm Markets and Markets. But experts and industry participants alike have expressed concerns about advanced A.I. replacing human workers.
As of early September, the $24.8 billion financial services giant had 1,800 active digital workers automating 1,500 tasks across more than 11 divisions, some of which have newly introduced generative A.I. capabilities, according to Brian Halpin, senior vice president of Internal Automation at SS&C. The company defines a digital worker as “unattended bots running in the back office.” The bots are powered by robotic process automation (RPA) technology—a rudimentary version of A.I. agents—that automates manual, repetitive tasks through predefined rules. The bots were introduced in January 2023 after SS&C acquired Blue Prism, a British company that makes RPA software.
SS&C estimates Blue Prism’s A.I. bots have resulted in over $100 million in “headcount savings” equivalent to full-time salaries over the entire 2023. Those savings came from open positions the company didn’t fill and later eliminated, according to Halpin. SS&C anticipates those savings to grow as it integrates more large language model capabilities into its digital staff. (The company employs 27,000 people globally, according to its website.) “It’s no different than having an offshore team,” Halpin, who leads the company’s efforts to deploy digital workers across the business, told Observer.
Generative A.I. is replacing “junior associates”
SS&C provides a range of custom software, including platforms that manage portfolios and loans for multinational banks, investment funds, insurance companies and other financial institutions. Traditionally, SS&C relied on manual labor for fund administration, outsourcing services, and back-office operations. Now, digital workers are deployed to automate rote tasks across every part of the business.
In SS&C’s business processing outsourcing divisions, for example, digital workers automate data-heavy tasks like stock transfers, account creation, loan processing and reconciliation—high-volume transactions on behalf of thousands of clients, Halpin said. SS&C’s transfer agency business uses digital workers to process client paperwork like shareholder reports for mutual fund record keeping. Digital workers even automate administrative tasks across SS&C’s internal divisions, such as human resources, IT and finance.
With the newly added generative A.I. capabilities, some digital workers can now process “complex documents” through prompts. For instance, a digital worker can process a credit loan agreement and extract specific data from that contract through textual queries like “What’s the close date of this contract?” or “What are the counterparties?”
“Repetitive tasks that require little thinking are where we see the great opportunity to provide the thinking worker more valuable outputs,” Rob Stone, senior vice president and general manager of SS&C’s Intelligent Automation division, told Observer. “Digital workers never get tired, never get frustrated and never get bored.”
As a result, tasks that once took minutes to complete can now be done in seconds, freeing up time for human employees to work on complex, creative tasks, according to Stone. The work digital workers do is comparable to the manual tasks “junior associates” take on across departments like operations, loan underwriting, and accounting, including “calculations” and “spreadsheet manipulation.”
That doesn’t mean humans are unnecessary for entry-level roles. The $100 million in cost savings from digital workers—more than a quarter of SS&C’s $385 million total operating expenses in 2023—were less about job elimination and more about increasing the existing workforce’s “capacity”, Halpin said. In other words, a team of 20 junior and senior level workers could now do the work of up to 50 people with help from digital workers. That could be, in part, why SS&C’s revenue grew 4.2 percent to $5.28 billion in 2023 compared to the prior year.
“You’re just giving them more support and more capability than they currently have,” Halpin said regarding SS&C’s human workers.
The rise of digital workers could have unintended consequences
Digital workers could set a precedent for employers to let go of lower-ranked employees down the line, said Tom Davenport, a professor of IT management at Babson College who studies A.I.’s impact on the workforce. Employees who do “repetitive, bacc-office” work—especially those outsourced to countries like India and the Philippines—are most vulnerable to A.I. displacement as the tech advances, Davenport told Observer.
“A.I. and automation are bad for entry-level workers,” Davenport told Observer. He used Klarna, a Swedish fintech company planning to replace thousands of its employees with A.I. over the upcoming years, as an example of what could happen if companies decide to integrate A.I. into their workflows. “It’s pretty clear that the long-term direction is to employ a lot fewer people,” the professor said.
Moreover, A.I. bots are prone to errors, which could lead to inaccurate outputs that degrade the quality of work. SS&C claims its digital workers don’t make mistakes, though if they did, operational managers who created the bots would be held accountable.
Training workers to learn how to manage and work with A.I. is what will keep humans safe from replacement, Davenport said. Since introducing digital workers, SS&C said it has been ramping up its A.I. upskilling efforts across hundreds of automation experts and division employees to configure and manage the bots.
Looking ahead, SS&C is on track to deploy up to 3,000 digital workers by the end of 2024, many of which will have additional generative A.I. capabilities, according to Halpin. As the digital workforce expands, SS&C will continue to train its human employees to work with digital workers, emphasizing that the bots will help employees get more done rather than replace them.
“There is just so much work that we still do that can be automated,” Halpin said.
