AB-410 Exam Questions

Semantic Kernel is the orchestration SDK for building AI agents that can “think” through an LLM and “do” through Plugins/Functions.

Implicit sharing lets the app creator share the AI capability without giving users direct access to the service keys.

PII, or Personally Identifiable Information, detection finds and masks sensitive data points in a text string.

The Alpha parameter controls the degree to which the search engine uses semantic vector scores compared with traditional keyword scores.

Entity Extraction is designed to extract structured data elements, or entities, from unstructured natural language text.

Security Roles together with Business Units define horizontal data access boundaries in an organization.

The Retrieval Engine searches the vector store to locate the most relevant context for inclusion in the AI prompt.

The system message supplies the AI’s core instructions, defining its identity and boundaries.

Zero-shot prompting relies on the model‘s internal general knowledge to satisfy a request without task-specific guidance.

Object Detection is built to recognize and locate specific visual entities and their counts within an image or video frame.

Power Automate serves as the logic engine that enables an agent to complete multi-step actions across different connectors.

Sentiment Analysis assesses text to identify emotional tone, which is important for escalation logic in support bots.

Max Tokens enforces a hard generation-length limit, which is important for controlling output size and token consumption costs.

The Unstructured Document model is intended for documents where text and data points appear in different locations.

Azure AI Studio includes automated evaluation tools for calculating metrics such as groundedness, coherence, and relevance.

Max Tokens sets a fixed limit on how many tokens are generated, directly controlling the physical length of the output.

This trigger supports global logic, such as data logging, PII masking, or text pre-processing, each time a user sends a message.

CLU is the specific service used to map natural language utterances to defined business intents or actions.

Augmentation is the stage where retrieved facts are added to the user’s original query, producing a context-rich prompt for the LLM.

Dataverse Search provides intelligent cross-table and fuzzy matching capabilities that are needed for robust grounding.

Higher temperature leads to more creative or random output, while lower temperature near 0 makes the model more deterministic.

The Azure AI Studio Playground enables developers to iteratively test, compare, and refine prompts across different models.

Copilot Studio lets you add SharePoint or OneDrive as a Knowledge Source so generative answers can be produced from your files.

Max Tokens applies a firm cap to the length of the generated response, which limits both the cost and the time needed for the model output.

Private Endpoints enable access to Azure AI services through a private IP address inside your Virtual Network (VNet).

Power Automate serves as the logic engine that enables an agent to perform multi-step actions across different connectors.

Top K is the number of the most relevant context items retrieved from the index and provided to the LLM for grounding.

Power Automate flows provide the bridge between Copilot Studio and external systems through standard or custom connectors.

A higher temperature produces more creative or random output, while a lower temperature near 0 makes the model deterministic.

AI Enrichment in Azure AI Search uses skillsets during ingestion to convert raw data into searchable insights.

Fairness means AI systems should treat all people equitably and should not perpetuate or introduce social bias.

Microsoft Purview supports data governance and auditing so organizations can track how sensitive data is used in AI applications.

Reducing temperature toward 0 makes the model focus more strongly on the most probable tokens.

Vector search uses embeddings to retrieve information by semantic meaning and context, even when the query uses different keywords.

Vector search finds semantic matches by using the distance between high-dimensional numerical text representations.

Azure AI Studio includes evaluation tools that calculate metrics such as groundedness, coherence, and relevance.

Custom Connectors let the Power Platform wrap REST or SOAP APIs, creating a bridge that allows Copilot to perform actions in legacy systems.

A temperature of 0.0 makes the model deterministic, producing consistent and focused responses by choosing the most likely next token.

Few-shot prompting gives examples that guide the model’s logic and formatting, producing more reliable outputs for specific tasks.

Few-shot prompting gives the model context through 2–5 examples, helping it understand complex or highly specific output requirements.

AI Builder recognizes the visual object, while Dataverse stores the metadata and links for the related manuals.

The System Message is where the AI model‘s tone, persona, and style constraints are defined.

Solutions are the standard way to package and move Power Platform assets through the Dev/Test/Prod pipeline.

Prompt Flow in Azure AI Studio lets developers build executable graphs that connect LLMs, prompts, and Python code.

Azure Monitor supplies telemetry, logs, and alerts for Azure AI services, supporting operational reliability and performance tracking.

Custom Connectors enable Power Platform tools to communicate with any service that exposes a RESTful API.

Global trigger phrases allow specific intents to interrupt and redirect the conversation flow from anywhere in the session.

Transparency makes the logic and data behind AI decisions clear and understandable to stakeholders.

The System Message defines the AI session’s core behavior, persona, and linguistic constraints, such as language.

Hallucinations happen when a model produces plausible but false information that is not supported by its training.

Azure APIM lets you secure, throttle, and monitor the APIs consumed by intelligent applications and plugins.

Accuracy is a standard metric for measuring how often the model’s predicted response matches the ground truth.

Power Automate flows work as the skills that enable an agent to take actions across more than 1,000 available connectors.

Zero-shot prompting depends on the model‘s internal general knowledge to complete a request without task-specific guidance.

Azure AI Studio‘s Model Catalog is the central location for discovering and deploying both frontier and open-source models.

Variables function as the bot’s short-term memory, enabling it to remember information the user provided earlier in the session.

Top P restricts the model’s word choices to the most likely candidates until their combined probability reaches P, supporting high-quality variation.

Reliability and Safety emphasize consistent performance under pressure and graceful handling of errors.

Unstructured models are layout-agnostic and use deep learning to locate fields regardless of where they appear physically in the document.

The embedding model transforms text into numerical vectors so similarity matching can be performed against indexed data.