Dis correct — includingwell-designed examplesin prompts is a key part offew-shot learning, which helps LLM-based agents better understand thetask structure, output style, and expected behavior.

UiPath encourages the use of examples for:

Classification(e.g., labeling sentiment, email categories)

Transformation tasks(e.g., turning unstructured text into tables)

Step-by-step instructions(e.g., troubleshooting flows)

These examples serve two purposes:

Pattern induction: The model picks up on consistent structures or rules used across examples.

Generalization: With diverse examples, the agent can apply logic to unseen but similar cases.

Best practice:

Usetypical, real-world examplesrepresentative of the data the agent will encounter.

Keep formatsclear and consistentacross input-output pairs.

Pair examples withexplicit instructionsin the system or user prompt.

Option A is flawed — focusing only on edge cases can confuse the model.

B is false — omitting examples forces the LLM to guess the structure, reducing accuracy.

C is misleading — examples improve performance butdo not guarantee perfect output; testing and evaluation are still required.

In short,prompt engineering with examples is essentialto buildingreliable, generalizable, and scalableAI agents.

The correct answer isA— zero-shot prompting refers toasking an LLM to perform a task without providing any prior examples in the prompt. In UiPath Agentic Automation, this is considered the simplest form of task prompting and is often used when:

The request isstraightforwardorfamiliar to the LLM

There'sno need for detailed contextor task demonstration

You want rapid generation without lengthy prompt design

UiPath distinguisheszero-shot,few-shot, andchain-of-thought promptingas part of itsPrompt Engineering Toolkit. While zero-shot is fast and scalable, it’s not ideal fornuanced or ambiguous tasks, which often benefit fromfew-shot examplesor structured reasoning steps.

Option B is misleading — complex scenarios usuallyrequiremore grounding.

C contradicts the definition of zero-shot.

D confuses prompting withmodel fine-tuning, which is a separate concept.

Zero-shot works well for common, templated tasks (e.g., classifying "Is this urgent?") but is less reliable in dynamic, multi-intent agent behaviors.

Cis correct — according to UiPath documentation,key-value pairs stored in Agent Memorypersist for12 months by default.

Agent Memoryis a persistent storage layer allowing agents to:

Recall decisions or context across runs

Store user preferences, status, or temporary flags

Maintain statefulness without relying on external databases

This capability is especially useful for:

Omnichannel customer interactions

Preference-aware recommendations

Tracking previously taken actions for continuity

Although memory storage is long-lasting (12 months), developers can:

Manually resetor expire entries

Use different memory scopes (e.g., per-user, per-agent)

Design memory-aware flows for personalization

Option D is incorrect — memory isnot auto-cleared on version updates.

A and B understate the retention policy — default expiration is clearly documented as12 monthsunless changed manually.

Agent Memory is a powerful enabler ofcontext-rich, stateful automations, especially for conversational or ongoing interactions.

Bis correct — the bpmn.uipath.com canvas is alightweight sandbox environmentfordrafting and visualizing agentic processes, butdoes not include full implementation capabilities. It is part of UiPath’s broaderMaestro experience, designed forearly-stage discovery, collaboration, and ideation.

Key characteristics:

Drag-and-dropBPMN modeling

Ability tooutline agents, decisions, automations, escalations

Useful forcollaborating with stakeholdersbefore technical development begins

Lacksdirect execution, tool integration, or runtime support

It is not a replacement forStudio WeborAutomation Cloud, which are used for:

Full implementation

Connecting to tools, prompts, or systems

Deployment and testing

Option A is incorrect — implementation requires transition intoStudio Web.

C is false — the tool is formodeling, not template import/export.

D misrepresents its role — it'snot the full-featured modeling tool, but adiscovery-phase sandbox.

Best practice: use bpmn.uipath.com todesign collaboratively, then export or map the flow inton8n,Studio, orMaestro production canvasfor build-out and testing.

Cis correct — agents in UiPath canintelligently automate complex procurement workflowsby combining tools likedocument extraction,web search for price benchmarks,policy validation, andLLM-based reasoning.

In this use case:

The agent extractsstructured data(item, price, quantity) from multiple quotations

Compares prices withexternal market sourcesusingWeb Searchor integrated APIs

Appliescompany policies or thresholdsusing system prompts and guardrails

Flags anomalies, escalates exceptions, or provides summarized comparisons

This reduces:

Manual effort

Human error

Turnaround time for approvals

And increases:

Policy compliance

Market alignment

Decision speed for procurement officers

Options A, B, and D all fall short of UiPath agent capabilities. These responses describepassive or limitedautomations, whereas agents are built to operateproactively and contextually, especially in high-value business functions like procurement.

This example reflects theagentic automation blueprintat work — combining perception, decision, and action across multiple systems in real time.

Bis the correct answer — in UiPath's Agent Builder (Studio Web),descriptions for input and output arguments serve as grounding contextfor the agent. These descriptions help the LLMunderstand what each argument represents, how it should be used in the generation process, and how to structure its outputs.

This is especially critical for:

Inputs like {{CUSTOMER_ISSUE}} — the agent needs to know it’s a complaint, question, or error

Outputs like {{TROUBLESHOOTING_STEPS}} — the agent should format these as steps, not just a summary

These descriptions:

Improve theaccuracy of prompt generation

Ensure the agentreturns structured, expected data

Help guide LLM behavior in multi-step or dynamic workflows

Option A is incorrect — Orchestrator triggers donot auto-mapbased on descriptions.

C is false — descriptions donot make arguments mandatory.

D is incorrect —output arguments benefit greatly from descriptions, especially for guiding LLMs on return format and content.

Bis correct — in UiPath’s agent framework,guardrailsplay a critical role incontrolling tool behavior and decision outcomesduring agent execution. Specifically, guardrails enable developers tohandle edge cases and define conditionsunder which:

The agent shouldescalate to a human

A tool should be skipped, modified, or retried

Output should be checked against validation rules

Guardrails workdeterministically, meaning they arerule-based conditionsapplied before, during, or after a tool runs — depending on the configuration. This allows for predictable and governed responses, such as:

“If tool output confidence is below 70%, escalate the task to Action Center.”

Option A is incorrect because guardrailscan and often do trigger human intervention.

Option C is false — guardrails can influencepre-execution, such as preventing tool calls under certain input conditions.

Option D downplays runtime functionality — guardrails are especially powerful during execution to protect against invalid results, failed API calls, or LLM drift.

UiPath promotes the use ofguardrailsto ensuresafe, accurate, and context-aware agent behavior, especially in regulated or sensitive environments.