Automation Anywhere buys its own AI agent vision

Automation Anywhere, a two-decade-old process automation vendor, has embraced the use of AI agents internally, in part to demonstrate the power of the emerging technology to potential customers.

But by launching more than 40 AI agents within its own systems since February 2024, the company has also saved several hundred thousand dollars by using agents to streamline financial processes, tech support, and other functions.

It was important for Automation Anywhere to “drink its own champagne” and advance traditional process automation with AI agents, but the company also saw a huge opportunity for itself, says Kapil Vyas, SVP for IT there.

“When generative AI really created this flash, in late 2022, we really saw this as something that we’ve been waiting for a very long time,” he says. “We realized that AI and automation was a marriage made in heaven — this is where the brains and the hands and feet come together.”

Agentic process automation

Automation Anywhere has moved from robotic process automation and intelligent process automation to what it calls agentic process automation, a framework that orchestrates teams of bots, AI tools, agents, traditional automation, and people, Vyas says.

The company’s use of internal AI agents, through a project codenamed “Putting AI Agents to Work,” has spread to several areas, with early success in its finance, tech support, and marketing departments.

Within finance operations, the company rolled out 12 targeted use cases that optimized processes across its order to cash (O2C), record to report (R2R), and tax operations. Across all finance functions, the AI agents led to cost savings of about $350,000, with about 6,000 hours of increased productivity. The agents also improved cash flow and mitigated risk to the tune of nearly $5 million.

In the O2C functions, for example, the company optimized billing and accounts receivable processes, improving its collection time and automating several manual processes.

“The idea was really to accelerate our financial flows,” Vyas says. “With financial processes like order to cash and record to report, they’re all bottlenecked by manual reconciliation and data validation.”

The AI agents help the company improve accuracy in financial processes and free up key finance professionals from repetitive work, allowing them to focus on analysis instead of data preparation, he says.

Within tech support, the company automated many incident resolution processes by deploying a self-service conversational agent for L1 support, ultimately saving about 33,000 work hours per year. A second agent-based process offers L2 and L3 support.

The support agents led to an 89% faster ticket resolution rate, and 30% of support tickets are now resolved autonomously, allowing the company to redeploy tech support employees to other functions, including customer satisfaction efforts.

The company receives nearly 10,000 tech support tickets a year, and the L1 support agent will handle more than a third of those requests. The L2 support agent automatically resolves midlevel complexity issue, and the L3 support agent leverages Microsoft Copilot to assist with complex issues.

By using tech support agents, “we’re always on 24/7 by 365,” Vyas says. “Employees get that instant gratification, and we were able to solve every problem about the traditional help desk.”

Finally, Automation Anywhere has used AI agents in its marketing efforts to generate new content, including blogs, videos, and social media posts, resulting in a tripling of materials created from the previous year, with cost savings of about 80%.

The agents are trained in the company’s unique brand voice and its style guide and are subject to a robust review process combining human and AI oversight. With productivity savings of nearly 15,000 hours per year, the agents have allowed the marketing team to focus on strategic initiatives.

Change management challenges bested

Automation Anywhere faced a handful of challenges when deploying its Putting AI Agents to Work project, which earned the company a 2025 CIO 100 Award for IT innovation and leadership.

Even though the company has a long history in automation, there was some employee resistance to change, along with a lack of AI skills in some departments, Vyas says. Business-focused employees, in particular, lacked AI skills, limiting their ability to reimagine agent-powered business processes. But as trust in AI grew, business users began designing processes that were fully autonomous.

In addition, there was a lack of trust in AI agent results among some employees, and the company focused on keeping humans in the loop during early agent efforts, then gradually removing human oversight in situations with limited risk, he adds.

Employee trust also touches on employment stability, with the company focusing on upskilling to keep workers relevant, Vyas says. “We make sure that they know this is not to take somebody’s job,” he says. “Everybody understands that they need to look at how is AI going to impact them.”

Agent-first mindset

While the company still has many human-led manual processes, it has focused on an AI- and agent-first mindset, with agents involved in decision-making at nearly every level, Vyas says.

“Agents will advise us to make better and quicker business decisions,” he adds. “Agents will guide us with defining strategy in real-time.”

To drive forward this agent-first mentality, Automation Anywhere has invested heavily in employee AI training, literacy, and trust, both through educational programs and hands-on experiences, Vyas says. The company also launched an Olympics-style competition that encouraged each of 15 teams to explore AI and automation by developing solutions to real business challenges.

“After we have that level of literacy, empowerment, and standardization, then building adoption,” Vyas says. “Now you are building momentum.”

The Automation Anywhere decision to deploy its own agents also builds momentum because it will help the company build better AI tools, says Arthur Villa, a research director focused on RPA and hyperautomation at Gartner.

“As an innovator, it’s a critical move because they need to do testing, they need to work out the bugs, they need to validate the usability within their own organization,” he says. “It gives them a chance to showcase their capabilities.”

The company, by deploying its own agents beginning in early 2024, has a first-mover advantage over many of its larger competitors, Villa adds.

“They have the advantage that they’re already on their second or third version of their agentic platform, while others just working out the kinks in their first,” he says. “In order for them to really capitalize on agentic, they need to push that narrative of, ‘We came here first, and we have all these agents in production, both internally and within our customer base.’”