4 min Applications

UiPath wants to harness AI agents with Maestro layer

UiPath wants to harness AI agents with Maestro layer

UiPath is making the building and organizing of AI agents easier than ever before thanks to a new Maestro orchestration layer inside the UiPath Platform. Given the company’s experience in process automation, it sees itself as the leading player in combining agents with security, compliance, and reliability requirements.

In fact, process automation has come under pressure from AI agents. The question is whether this form of AI could replace or incorporate RPA. However, UiPath wants to keep agents in line in the same way as process automation, making its own solution indispensable.

Mark Geene, senior vice president and general manager of AI products at UiPath, emphasizes that the focus on processes limits the number of vendors that can orchestrate end-to-end processes. The company’s orchestration layer, called Maestro, can dynamically switch between agents, robots, and people to execute long-running workflows.

Intelligent orchestration

Maestro is described as a tool that automates, models, and optimizes complex business processes with built-in process intelligence and real-time KPI monitoring. This technology enables continuous optimization of large groups of agents. The solution builds on the process intelligence that was already part of UiPath’s RPA offering.

Process intelligence, according to Geene, “understands how a process runs, where the bottlenecks are and how the process is routed. We find inefficiencies and make recommendations. It’s not a static workflow, but can adapt during execution.” UiPath provides a set of predefined KPIs, while customers can also create their own indicators.

It is far from the only player with an idea about orchestrating AI agents. ServiceNow is also claiming a central position, positioning its platform as a control tower for AI agents. Like UiPath, ServiceNow focuses on raising the level of existing workflows by making them available to AI agents. The same applies to Salesforce with Agentforce and IBM, which sees itself as an excellent candidate for keeping AI agents organized.

Built-in governance

An important part of the new UiPath platform is built-in governance. Maestro ensures that AI agents operate within clearly defined security frameworks with predictable performance inside UiPath’s platform. The platform offers real-time vulnerability assessments and data access controls. Developers can build prototype agents in UiPath Studio using low-code tools and the Python programming language.

According to Geene, rules can be defined for both the agents themselves and at the orchestration level. Agents are given a goal, with context and policy. Geene refers to built-in escalations, which ensure that people remain involved if the agent is not entirely sure what to do.

Integration with existing systems

UiPath says the platform also integrates with and supports agents built with third-party frameworks, including LangChain, Anthropic, and Microsoft. It can also automatically create orchestrations from processes defined in Business Process Model and Notation, a standardized graphical notation for business processes.

For document processing, UiPath has introduced a new feature called Intelligent Xtraction & Processing. This uses multi-modal, AI-based classification and extraction of unstructured data for applications such as claims allocation, loan initiation, and electronic batch records.

The UI Agent for computer use, currently in private preview, is a natural language-driven agent that can actually use a PC as if it were a real user, allowing it to be flexibly deployed for a variety of purposes.

Learning curve and applications

Geene acknowledges that there is a learning curve for developers familiar with traditional RPA. He points out that traditional automation is a step-by-step process. Agents, on the other hand, become more autonomous because they are not given a fully structured workflow. This means that they are not always the first choice for a particular task.

As with other AI agent platforms, adoption is likely to start with specific use cases where the benefits are most apparent. For many organizations, this will require a step-by-step approach, first getting the data architecture and knowledge base in order before they can fully benefit from AI agents.

UiPath’s agentic automation platform is delivered from the company’s cloud and requires no user infrastructure. Pricing is based on the frequency of workload execution and the number of tokens required. This is in line with the trend toward cloud-based AI solutions that offer flexible cost models.

With the introduction of this platform, UiPath is clearly positioning itself in the emerging market of agentic AI, where it competes with players such as Google’s Vertex AI and IBM’s watsonx Orchestrate. The question now is which players in this field will become the dominant orchestration platforms for managing and optimizing AI agents. The winner will likely only emerge in several months or even a few years, once the value of agentic AI can truly be realized.