In 2003, Mihir Shukla had just navigated the late 1990s tech boom having been involved in early iterations of two pioneers of internet search: Netscape and Infoseek. Trained as a software engineer, he also had experience in mobile technology. Shukla seemed to have a knack for identifying early-stage technologies with the potential for explosive growth. So, when he was looking to start his own company, he leaned on that intuition to guide him.
Automation Anywhere is a trailblazer in the development of business process automation software. Like his other ventures, Shukla chose a nascent industry when looking for opportunity. In the two decades since, the technology Automation Anywhere started with and helped pioneer—robotic process automation (RPA)—has given way to its leadership of a new category it created, agentic process automation (APA). Shukla has watched as automation platforms have grown to encompass not only RPA, but other advanced technologies including natural language processing, intelligent document processing, machine learning, generative AI and, most recently, agentic AI.
Automation Today sat down with Shukla recently to revisit what motivated him to build his company and what advancing technology means for the process automation space.
“In the early 2000s I had completed five multi-billion-dollar journeys in my professional career, and so when time came to do the sixth one, I needed a good reason,” Shukla remembers. “Computers transformed my life, so I wanted to pass the same kind of transformation on to the next generation. I decided to create a piece of technology that would take automation beyond where computers had brought it to that point.”
He said co-founding Automation Anywhere was an attempt to chase a larger vision to reallocate the intellectual capacity of the planet. The power of the world’s knowledge workers was being applied to unimaginative tasks, he thought. Imagine if that capacity could be directed at work that mattered more.
“Rather than putting our focus on processing invoices or claims, if we aim the human intellectual capacity in the right direction, we can solve the most challenging problems of our times,” Shukla says. “When that capacity is focused on more interesting problems, amazing things happen.”
And, while amazing things have been happening in the automation space for more than two decades, the past two-plus years have been especially exciting. Since the public unveiling of Chat GPT in November of 2022, first generative AI and then its more advanced cousin, agentic AI, have dominated the industry. Automation platforms have integrated both technologies, supercharging their capabilities and touching off an arms race among providers.
According to Shukla, Automation Anywhere aims to be at the vanguard of that race. He is not only a believer in agentic AI as an important component of a comprehensive intelligent automation platform, but he’s also interested in the next steps as the technology advances even farther.
Automation Anywhere has a vision to deliver the Autonomous Enterprise, which means, Shukla says, up to 80 percent of business processes and tasks are made fully autonomous or AI-assisted. It has already started executing on this vision with its most advanced customers who are seeing how workflows can be managed by APA with AI agents that reason, adapt and orchestrate work in real time, department by department.
In his conversation with Automation Today, however, he stressed that providers must overcome the trust issues many users have with all forms of AI.
“From a purely technological perspective, it’s relatively simple to get end users to understand the value of agentic AI,” Shukla says. “I would ask them what business problem they are trying to solve.”
He relays a conversation he had with an aircraft manufacturer trying to account for new tariffs in a business where their product comprises 100,000 parts sourced from many countries.
“This is a problem that will simply take too long for humans to solve,” he explains. “Traditional intelligent automation platforms could use RPA, intelligent document processing and APIs that can communicate with multiple systems to get an answer. But when you add agentic AI to the mix, you are capable of so much more. AI agents can look at a vast amount of data in the context of a given environment and guide you to the most likely answer. In this example, say one of the parts is not available. An agent designed for this purpose could identify a replacement part from a different supplier and make the purchase.”
But, if users don’t trust the tech, they won’t invest in it and they won’t use it. Shukla understands that providers must strive for demonstrable security to engender the confidence to move forward with agentic AI projects.
“The Automation Anywhere platform is a security- and trust-first architecture, which means every single step and every aspect of the platform has a guardrail,” he notes. “We have built in observability and generate a confidence score for every single step, so users have full understanding of what the process is, how it is running, and how confident it is that it’s making the right decision.”
In the end, Shukla says, humans can use the confidence score to assess the risk and decide whether to move ahead with the automation.
People are always concerned about new technologies, he says, and the burden of proof is on providers to demonstrate repeatedly that they work, they’re secure and they’re accurate. Seeing is believing, Shukla says, and being able to prove the platform works has led to success for Automation Anywhere.
“At the beginning of this conversation, we said the answer is to solve customers’ problems,” he says. “It turns out the best way to solve problems is to combine multiple technologies and orchestrate them effectively. We’ve shown customers how to do it, and they could see it working. They were able to pick the right automation technologies for the right problems and orchestrate all of it to solve those problems. We have about 1,500 AI agent deployments and seeing is believing. When people see a piece of technology versus everybody else just talking about it, they became increasingly convinced that this is how problems get solved using our agentic process automation system.”
