Integrare AI e Automazione con UiPath Agent Builder

Unisciti a noi in questo webinar dove esploreremo le funzionalità di Agent Builder e come sfruttare l'AI nel mondo dell'automazione.
Scopri come si possa aumentare l'efficienza dei processi aziendali in ambito Finance, HR e Manufacturing attraverso esempi pratici e casi d'uso reali.
Inoltre sarà possibile interagire con altri membri della Community e condividere opinioni e punti di vista sulle nuove tecnologie in ambito AI.

1. The UiPath word mark, logos, and robots are registered trademarks owned by UiPath, Inc. and its affiliates. UiPath (R) is a registered trademark in the United
States and several countries across the globe. See TMEP 906. ©2025 UiPath. All rights reserved.
Integrare AI e
Automazione con UiPath
Agent Builder

2. 2
Agenda slide
01
02
03
What is the Agent Builder
Agent Builder Components
Build your Agent

3. 3

4. What are Agents?
4
Agents are the next evolution in automation. Powered by technologies like large
language models (LLMs), machine learning, and traditional enterprise automation,
agents are designed to operate in dynamic, non-deterministic environments. They can
plan, act, learn, and adapt—making them ideal for processes that require judgment,
flexibility, and contextual awareness.
4
Agentic automation is the orchestration of agents, robots, and humans in a unified
automation ecosystem. On the UiPath Platform, agents of all types can operate with
built-in guardrails, governance, and security—making them enterprise-ready from day
one.

5. What are Agents?
5
5

6. How to create an Agent
6
6
From prebuiltagent
Asetoftemplatesis
availableforreutilizationin
StudioWeboraprojectbuilt
bysomecollaborator
Autopilot
AI feature embedded
in Studio Web, allowing you
to create an agent starting
from a prompt
Build from scatch
Using Studio Web and
creating a new Solution

7. Autopilot
7
7
➢Autopilot automatically creates input and output arguments (and their structure) based on
your description. E.g., the input arguments capture exactly what the users need to provide
(city, event type, seating preference, participant count), and the output argument(s) defines the
agent's response and its format (item arrays with names, descriptions, contact details, pricing of
venues).
➢Autopilot crafts system prompts that guide the agent's behavior by defining the search logic,
specifying response requirements along with a fallback mechanism ('Contact venue for pricing’).
➢It crafts the user prompt that translates business logic into natural interactions by adding input
validation and capacity checking.
➢Autopilot can choose the right tools for your agent's goal. E.g., the web search tool finds
venues that match your criteria across the internet. Similarly, the web summary tool extracts
detailed information from venue websites

8. Agent Builder Components
8
System Prompt
A system prompt is a foundational instruction set that guides the behaviourof an AI
agent. It defines the agent’s role, objectives, constraints, establishes rules for decision-
making, and provides instructions on when to use specific tools, reference contextual
information and escalate issues
8
User Prompt
User prompts are specific instructions, requests, or questions a user provides to an
agent during an interaction. It clearly outlines the task to be performed, includes any
necessary context or input data, and specifies the desired output format.

9. Agent Builder Components
9
Tools
Applications that can be used by the agent to analyze and furtherly evaluate the prompt.
•RPA Workflows
•API Workflows
•Other Agents
•Connected services
9
Contexts
Data storage to a dedicated space within UiPath Orchestrator, thereby providing a
centralized and secure way to store contextual data used across multipleagentic
automationsolutions. Indexing the data can help the agent identify the request and
extrapolate the relevant data in a faster and more reliable way
Escalations
Escalationsdefine when and how your agent should hand off tasks to human workers,
ensuring that critical decisions and complex scenarios receive appropriate human
oversight.

10. Agent Builder Components
10
While choosing your grounding source always ask: "Is this information dynamic or
static?" Use connectors for frequently changing content (like policies or documentation
maintained by business users) and storage buckets for stable, version-controlled files
like templates or service mappings. The right choice improves reliability and reduces
maintenance overhead.
Before creating an escalation, ask: "What situations truly require human
intervention?" Not every agent decision needs a manual check. Reserve escalations for
moments where user intent is unclear, data is missing, or business risk is high. This
keeps the human-in-the-loop experience efficient and meaningful.
10
Best Practices

12. https://community.uipath.com/milan/
[email protected]