What I like most is how easy it is to build AI agents without extensive coding.

The low-code experience feels familiar for teams already using UiPath.

It allows us to combine AI reasoning with existing automation workflows.

Testing and refining agents is straightforward.

This helped us experiment with agentic automation much faster. Review collected by and hosted on G2.com.

One area that could be improved is debugging and troubleshooting.

When an agent does not behave as expected, identifying the root cause can take time.

More detailed logs and diagnostics would be helpful.

Since the product is still evolving, some features feel less mature than core UiPath offerings.

Overall, these have been manageable challenges. Review collected by and hosted on G2.com.

What I like best about UiPath Agent Builder is how easy it is to build enterprise-grade AI agents while leveraging the full UiPath ecosystem. The low-code experience makes agent development accessible, and Autopilot significantly accelerates development by helping create agents from natural language descriptions, generating inputs and outputs automatically, refining prompts, and answering questions during the development process.

I also appreciate the built-in tools available to agents, such as Analyze File, Context Grounding, and Deep RAG capabilities, which make it easy to work with enterprise documents and unstructured data without building everything from scratch. Context Grounding allows agents to retrieve information from governed knowledge sources, improving response accuracy and reducing hallucinations.

Another major advantage is the flexibility to choose from multiple LLM providers and models, allowing us to select the most suitable model for each use case while maintaining enterprise governance. The platform's vendor-agnostic architecture provides flexibility without locking us into a single AI provider.

The human-in-the-loop capabilities are also excellent. Agent Builder integrates seamlessly with Action Center, allowing agents to escalate tasks for review, approvals, and exception handling whenever human judgment is required. This is especially valuable for compliance-focused processes.

One of the strongest differentiators is the ability to use existing UiPath automations, APIs, and RPA workflows as agent tools. This allows agents to not only reason and make decisions but also execute actions across enterprise systems using proven automation assets that are already in production.

Finally, deployment and lifecycle management are very straightforward. Agent Builder makes it easy to test, version, deploy, and manage agents, and they integrate seamlessly into Maestro-orchestrated processes and existing RPA workflows. This makes it simple to move from experimentation to production while maintaining governance, visibility, and scalability. Review collected by and hosted on G2.com.

At this stage, I have not encountered any major drawbacks with UiPath Agent Builder. Review collected by and hosted on G2.com.

What I like best about UiPath Agent Builder is how well it supports building AI agents for reasoning and decision-making tasks. I primarily use it to create compliance-focused agents, and it handles that kind of structured, rule-sensitive logic really well. The interface makes it straightforward to design and configure agents without a heavy setup process, so I can go from idea to a working agent quickly. The conversational AI assistant built into it is genuinely helpful when I'm refining how an agent should behave or make decisions. Configuring tools is refreshingly easy — wiring up the capabilities an agent needs takes very little effort, and prompt orchestration is intuitive enough that I can shape how the agent reasons and responds without fighting the interface. Agent versioning is another strong point: being able to track changes and roll between versions makes it much safer to iterate on agents that are tied to compliance and decision logic. Review collected by and hosted on G2.com.

Honestly there isn't much I dislike. But the Indexing part can be improved Review collected by and hosted on G2.com.

UiPath Agent Builder is a strong addition to the UiPath ecosystem, making it easier to design and deploy AI-driven agents that can handle complex, decision-based workflows.

What stands out to me is how it supports agents that go beyond rule-based automation by incorporating reasoning, context awareness, and multi-step execution. The interface makes it straightforward to set up agent logic quickly, and its integration with existing UiPath components (like Studio, Orchestrator, and AI Center) helps you embed agents smoothly into broader workflows.

In my experience, it’s been especially useful for working with unstructured inputs such as emails and documents, and for automating decision-heavy processes that were previously difficult to scale with traditional RPA alone. Being able to combine AI models with workflow orchestration has noticeably reduced the amount of manual intervention needed.

Another advantage is the speed from concept to implementation. Work that used to require custom AI pipelines can now be prototyped and iterated on much faster within the platform.

On the downside, designing and tuning agent behavior still takes experimentation, and I’d like more transparency and observability into how agents arrive at their decisions. There’s also a learning curve when you start working with more advanced agent configurations.

Overall, UiPath Agent Builder feels like a key enabler for moving toward agentic automation, helping organizations build smarter, more adaptive workflows. Review collected by and hosted on G2.com.

One challenge I’ve found with UiPath Agent Builder from UiPath is that designing and fine-tuning agent behavior often takes a lot of experimentation, particularly for more complex use cases.

I’d also like better visibility into how agents arrive at their decisions, since the current level of transparency can make debugging and optimization more difficult.

On top of that, there’s a learning curve when it comes to advanced configurations. Having more prebuilt templates and clearer best practices would help speed up adoption and make it easier to get consistent results. Review collected by and hosted on G2.com.

I like how easily it evolves from a raw prototype into production-ready agents. Review collected by and hosted on G2.com.

The cost is a concern, and the escalation rules feel too rigid. It also returns vague errors when LLM calls fail, which makes it difficult to trace the underlying reason for the error. Review collected by and hosted on G2.com.

It has a bit of a learning curve and looks slightly different when compared to traditional automation building, but it's very intuitive! I'm excited to use in a production environment. Review collected by and hosted on G2.com.

The UI feels cluttered at times with it being within the browser. Review collected by and hosted on G2.com.