The correct answer isB. This prompt follows UiPath'sbest practices for system promptsby clearly establishing agent identity, defining behavior logic, and including formatting constraints - all in a numbered, readable structure. The agent is given a clear role ("supplier invoices only"), boundary rules ("reject any other request"), and step-by-step instructions to follow. Numbered steps improve clarity and make parsing easier for LLMs.
The inclusion of tool usage (LookupInvoice) and conditional logic (# $10,000 vs > $10,000) mirrors UiPath's orchestration standards. Importantly, it also specifies how to format the output using <invoice_status> tags and instructs the agent to maintain a professional tone - critical elements in UiPath'sPrompt Engineering Framework.
Compared to options C and D, which introduce a rigid JSON format, Option B balancesstructure with flexibility. JSON-only prompts (like C) are good for strict APIs but lack the natural language behavior, tone control, and task-scoping essential in real-world agents. Option A is close but lacks step numbering, making it slightly less robust.
UiPath recommends system prompts include:
* Agent persona and role
* Tool instructions and decision rules
* Tone and refusal handling
* Clear, consistent output formatting
Option B satisfies all these criteria, making it the most robust, agent-ready system prompt.

The correct answer isC- UiPath'sImpact and Feasibility Matrixis a structured tool used in thediscovery and prioritizationphase of agentic automation. It enables teams toevaluate and rank automation opportunitiesbased on two key dimensions:
* Impact: Thebusiness valuedelivered - including time savings, risk reduction, efficiency, or user experience improvement.
* Feasibility: Howpracticalorcost-effectiveit is to implement - considering technical complexity, data availability, resource constraints, and integration readiness.
This matrix helps classify use cases into quadrants such as:
* Quick Wins(High Impact, High Feasibility)
* Strategic Bets(High Impact, Low Feasibility)
* Do Later(Low Impact, High Feasibility)
* Avoid or Backlog(Low Impact, Low Feasibility)
UiPath emphasizes that this method ensures teams focus efforts whereagentic automation can create real business value quickly- avoiding wasted time on low-priority or hard-to-execute ideas.
Options A and B are partial approaches that ignore one of the two axes.
D is incorrect - not all processes should be automated, especially if they're low-value or high-risk.
This balanced framework is a core part of UiPath'sAgentic Design Blueprintmethodology for aligning automation with strategic priorities.

Cis correct -deterministic evaluationsare best suited for cases where thecorrect output is known and fixed
, allowing for binary or rule-based validation.
Examples include:
* Exact matches (e.g., status: "Approved")
* Regex pattern checks
* Structured JSON outputs
* Correct field extraction (e.g., invoice number = INV-2023-0021)
UiPath supportsdeterministic evaluationusing logic like:
* "Output equals Expected"
* "Contains X and Y"
* "JSON schema is valid"
This is distinct frommodel-graded evaluations, which are used when outputs areopen-endedorqualitative(e.
g., summarization, sentiment, tone). These require LLM-based grading to assess whether the output is "good enough" even if it varies slightly.
Option A and B refer tosubjective assessmentsbetter suited formodel-graded scoring.
D implies feedback-driven quality, again requiringflexible interpretation, not deterministic checking.
Deterministic methods offerspeed, clarity, and automationin validation - ideal for tasks where there'sonly one right answer.

Bis correct - creating a newevaluation setin UiPath involves a multi-step process designed to enable qualitative and quantitative review of agent behavior.
Steps include:
* Namingthe evaluation set
* Addinginput promptsandexpected outputs
* Saving each test item (often called "evaluations")
* Assigning evaluators, who will manually or automatically score the results This process enablestestable, repeatable evaluationof agent behavior before deployment - ensuring the model produces correct, useful, and safe outputs.
Options A and C are incorrect:
* A reverses the order: inputs and expected outputs are neededbeforeevaluators.
* C is false - evaluation setscan be built from scratch.D implies scoring is automatic, but human reviewers or comparison logic are often required for nuanced evaluations.
This aligns with UiPath's best practices inagent validationand post-deployment assurance.

The correct answer isD, and this is a foundational concept in UiPath'sAgentic Discovery and Design Blueprint methodology.
Anagent storyserves as aclear, narrative-driven blueprintthat describes:
* What the agent does
* For whom it works
* When it activates
* How it makes decisions
* What success looks like
UiPath emphasizes that a well-crafted agent story ensures alignment betweenbusiness stakeholders,subject matter experts (SMEs), andtechnical developers. It keeps the development team focused on value delivery by outlining thecore capabilities,contextual behavior, andinteractionsof the agent in a human-readable form.
This approach is critical during thedesign phase, as it:
* Prevents scope creep
* Clarifies success metrics
* Enhances stakeholder buy-in
* Anchors prompt design, orchestration, and escalation logic
UiPath also uses the agent story to guidegrounding strategies, tool selection, and even escalation paths - making it much more than a documentation artifact.
Options A, B, and C misrepresent the function of agent stories. Only D captures its value in focusing the team onwhat matters most for delivering real business outcomes.