AI’s next act is agentic: It’s not just thinking—it’s doing

It’s hard to believe in hindsight, but when the internet and cloud computing burst onto the scene some three decades ago, their potential was largely underestimated. Today, we know that those who fully embraced these technologies—not merely as productivity tools, but as engines for radical business transformation—ended up building the largest, most powerful enterprises in history.

Many assume such a revolution won’t happen again. But history suggests otherwise—steam power and electricity were also initially underappreciated, yet each ushered in an era of unprecedented change.

Today, we stand at the dawn of another such technological transformation: agentic automation.

The transformative opportunities arising from agentic automation were the focus of our most recent quarterly webcast, UiPath Live: Rethinking Your Operating Model for the Agentic Age. Our guests—a world-renowned AI academician, an AI technology trends expert, a developer creating agentic automations, and UiPath Founder and CEO Daniel Dines—provided some important insights into the subject.

But before we dive into how agentic automation is poised to reshape business as we know it, let's take a step back and define what we're talking about.

What is agentic automation, anyway?

When UiPath Live hosts Mary Tetlow and Geoff Anderson put this question to UiPath Founder and CEO Daniel Dines, he explained that agentic automation is a fusion of AI, automation, and orchestration–a “basically, a set of technologies that help us deliver AI autonomously.” He added, “what's really transformational is when you can take a human activity and deliver it completely autonomously.”

Live guest Ariel Bernstein of Accenture asserts that agentic automation represents an evolutionary leap from robotic process automation—taking automation to a whole new level and dramatically increasing the surface area of automation in an enterprise. At the heart of this transformation: AI agents, who leverage a combination of agentic AI and automation to dynamically perceive their environment, analyze data, navigate changing workflows, and make decisions all on their own.

Agentic automation empowers AI agents to work independently of people to frame, plan, and direct the execution of complex tasks. Agents can size up a situation, figure out what needs to be done, and execute the necessary actions even in dynamic situations where a rules-based approach would falter.

What’s so great about it?

For organizations, this unlocks a brand new source of labor—a virtual workforce—that’s capable of independently taking on complex tasks. And this, in turn, opens up exciting vistas for different operating models, characterized by innovation, agility, and the full utilization of both internal and external data as well as a host of other benefits. Best-selling author and AI authority Tom Davenport took us through a number of them:

Better decision-making at warp speed

From real-time supply chain optimizations to detecting fraudulent transactions, businesses consistently face crucial decisions that rely on rapid analysis of vast amounts of data. AI agents, with their ability to process and analyze information at superhuman speeds, provide high-quality insights instantaneously, ensuring your decisions are always grounded in the most up-to-date and accurate data.

Happier, more engaged employees

Offloading a wider range of tedious processes to agents allows employees to spend more of their time and energy on creative, strategic work. This is a win-win—employees are both happier at work and able to make a more significant impact on their organization’s goals.

Continuous process improvement

The best agents continuously learn and adapt, becoming more effective over time and driving ongoing process improvements.

Five ways that enterprises can set themselves up for success in the agentic age

"Agentic AI is still in its infancy," remarked Mary Tetlow, Chief Brand Officer at UiPath, during the recent UiPath Live show. "Most people don't know what it is yet.”

This presents a golden opportunity for organizations—early adoption of this technology can give you a significant edge over competitors. Live guests provided some advice on how to do so:

1. Build on what you have

Even though agentic automation is brand new, most companies don’t have to start from scratch. “For larger enterprises, one of the biggest advantages they have is years and years of data and troves of processes…that they can feed these machines to make them more valuable,” noted Bernstein. Tom Davenport added that organizations with a clear understanding of their end-to-end processes will have an easier transition.

Moreover, you may already have some automated processes in place that can more easily morph into agentic automation. For example, if you use advanced intelligent document processing (IDP), you're already extracting meaning from documents. Agentic automation can take it a step further by automating the actions that follow.

UiPath Senior Technical Account Manager Kevin Bernstein—who has built agentic processes for both UiPath and our customers—provided an example: mortgage application processing. If a company is already using IDP to extract financial data from the applicant's documents, AI agents could take on the next steps currently performed by people, such as comparing the data with the bank’s lending criteria. The agent could then make an informed recommendation on mortgage approval for review by the human loan officer.

2. Create a sound agentic infrastructure

Even the smartest, most effective employees struggle to succeed in a chaotic workplace. The same is true for agents—they need a well-organized infrastructure to perform at their peak.

Enterprises will employ a variety of agents, ranging from specialized ones focused on specific tasks (like processing invoices) to more versatile agents capable of handling a broader range of activities. There's no one-size-fits-all "best" agent. The key to success lies in orchestration.

I wouldn't be able to point to one specific agent as being better than the other. It's going to be how they end up being deployed and, more importantly, how they're deployed as part of a larger ecosystem.”

Ariel Bernstein, Senior Manager, Accenture

Daniel put it simply: “Without orchestration, there is no agentic AI.” AI agents excel at executing tasks across systems, but this also increases the risk of their wires getting crossed. Add other players like robots and GenAI into the mix, and the need for a central coordinator becomes clear.

Orchestration provides the essential framework for directing the actions of multiple agents, ensuring they’re working together to achieve business goals. It provides central governance and enables human oversight, ensuring that agents have the appropriate level of autonomy—and people have the appropriate level of control.

3. Remember that the future of automation is agentic AND robotic

Agentic automation does not replace RPA. Rather, agents will draw on RPA robots to execute the tasks at which robots excel and which abound in any end-to-end work process: structured, rules-based, repetitive ones. Agents will need these efficient, accurate robotic resources more than ever. So think of agentic automation as the conductor and RPA as the orchestra—each plays a critical role in creating a symphony of seamless, end-to-end automation.

4. Look at the processes you haven’t been able to automate end-to-end

Agentic automation dramatically expands the surface area of automation in the enterprise to include complex, end-to-end processes that you may have wanted to fully automate—but were simply too complex and unpredictable for RPA to handle on its own. A good starting point might be to start with this long tail and see what’s newly possible.

Agentic automation may help you fill some of the stubborn gaps in a process, too. For example, a common problem stymying many companies is the need to split an address into its individual components. It’s a more complicated task than it seems, because it’s highly variable. While Kevin Bernstein says that, “in concept, [this process] is something you could have done with RPA,” agents make it a much less onerous task. “Agents can go further and it becomes simpler—less code, more context.”

5. Understand your processes and data

Maximizing the potential of agentic automation hinges on a rock-solid understanding of your organization’s data and processes. Kevin Bernstein says that this is where process mining and task mining come in—they shine a spotlight on the intricacies of your workflows, revealing bottlenecks, inefficiencies, and prime agentic automation opportunities.

Understanding your data is equally important. AI agents need access to accurate and relevant internal data to make informed decisions. Context grounding techniques equip agents with the necessary background and situational awareness, while a knowledge graph organizes and connects your data, helping agents understand the complex relationships within your information ecosystem.

The agentic AI age is dawning. Are you ready?

The way modern, successful companies operate five years from now will be almost unrecognizable. The agentic operating model will make them faster, more efficient, and more effective. Their exceptional customer engagement, adaptability to change, and empowered workforce will set them apart.

The time to act is now. Begin your agentic automation journey by starting small and gradually expanding agentic capabilities. Doing so will give you the knowledge and experience needed to scale up to larger and more impactful use cases before the competition.

We’ve just scratched the surface here about all there is to learn about agentic automation. To delve deeper and hear more from our distinguished guests, check out the on-demand recording of the entire "UiPath Live: Rethinking Your Operating Model for the Agentic Age."

Hear and see agentic in action at UiPath FORWARD + TechEd in Las Vegas.

There's still time to register.