Building a conversational agent is straightforward. Building one that actually works reliably in a regulated, enterprise environment is not. That requires a real process for design, testing, deployment, and monitoring.

This session walks through the complete conversational agent lifecycle using Agent Builder in UiPath Studio. You’ll see how to move from prototype to production with a structured, repeatable process - whether you’re a developer or a technical automation professional working in a low-code environment. You do not need to write custom code to follow this lifecycle.

We begin with an architectural framing: when is a conversational agent the right interaction model, and when is an autonomous or workflow-driven agent more appropriate? Making the right design choice upfront prevents unnecessary complexity and ensures alignment with business objectives.

From there, we move through each stage of the lifecycle:

• Design: Writing effective system prompts, choosing models, configuring tool types
• Evaluation: Working with debug chat and real-time execution traces, building test case libraries for single-turn and multi-turn conversations, using Autopilot to auto-generate trajectory-based evaluations, and validating behavior before you ship
• Deployment: Publishing through solution management, folder architecture and robot accounts, the production chat interface, and how parallel tool execution works with guardrails
• Observability: Performance dashboards, user feedback analysis, AI Trust Layer auditing for compliance, and tracking PII and data residency requirements

👨‍💻 Speakers:

Tuan Nguyen, Senior Manager Product Management UiPath