Free AB-410 Exam Flashcards

Analyze the requirement to identify which components (data model, app type, flow, agent) and implementation options actually fit it, rather than jumping straight to building.

Whether a built-in (prebuilt) agent already covers the scenario. The skill area explicitly calls for evaluating built-in agents before recommending custom agent development.

Choosing how to extend a solution beyond native Power Platform features, such as a plug-in, custom connector, or external Azure service, when out-of-the-box tools cannot fully meet a requirement.

ALM strategy is the plan for moving solution components through environments using solutions, environment variables, connection references, and pipelines. Environment strategy (Dev/Test/Prod separation) is one input to that broader ALM plan, not the whole plan itself.

Standard table. It ships with prebuilt schema, relationships, and business logic from Microsoft, while a custom table is reserved for schema a maker defines from scratch.

A prompt column. It calls an AI Hub/AI Builder prompt using data from the current record to populate the field automatically, rather than requiring manual entry or a same-row formula.

Table-level settings such as ownership type (user/team-owned vs organization-owned) and audit configuration, which apply to the whole table rather than a single field.

An AI-written natural-language overview of a single record's data, giving users a quick recap without opening or scrolling through every field on the form.

A global choice (option set) column definition. Defining the choice globally lets multiple tables and columns reuse the identical option list instead of redefining a local choice on each table.

N:N (many-to-many). It uses an intersect table behind the scenes, unlike a 1:N relationship, which is a simple parent-child lookup where the child holds one reference to the parent.

A public view is a shared, maker-defined query on a table that all users can select; a personal view is a filtered list an individual user creates and saves for themselves only.

A field security profile. It grants or denies access on individual secured columns, which is a separate control from a security role's table-level (row) privileges.

Hierarchy security. It automatically grants managers visibility into the records of people below them in the position or manager hierarchy, on top of normal security role access.

The row-level access scope for a given table, such as whether a user can see only their own records, their business unit's records, or the entire organization's records.

The site map. It is the app's navigation skeleton, separate from any individual form, view, or dashboard the app contains.

A generative page. It is an AI-created model-driven app page produced from a natural-language prompt instead of manual layout design.

A form lays out fields for creating or editing a single record; a view is a filtered, column-based list showing multiple records at once.

Configure access to that form or view so it is assigned only to the intended security roles, hiding it from roles that were not granted access.

A dashboard. It aggregates multiple charts and views together, whereas a chart alone is just one aggregated visualization of a single view's data.

Yes. A canvas app connects to a data source through any supported connector, including non-Dataverse sources, which is a key contrast with model-driven apps that are Dataverse-only.

A named formula. It is a reusable, app-scoped calculation defined once, similar to a global variable but driven by a live Power Fx formula instead of an assigned value.

A user-defined function. It wraps reusable Power Fx logic with defined parameters so it can be called consistently from multiple places in the app.

Package it into a component library. The library holds reusable components that any canvas app can import, keeping the control consistent across apps.

A delegation issue. Non-delegable functions only evaluate against the records already loaded locally instead of the whole data source, which hurts both performance and accuracy on large datasets.

A variable holds a single app-scoped value, while a collection is an in-memory table that can store, filter, and manipulate multiple rows of structured data.

IfError, often paired with Trace/logging. It intercepts a formula error and lets the app branch to fallback logic instead of surfacing the raw failure.

Monitor is the live diagnostic trace tool for real-time debugging during a session. Test Studio is different: it runs saved, automated UI test scripts rather than showing a live trace.

Create a Copilot Studio agent directly from the canvas app. The designer can generate an agent that plugs into the app's context instead of requiring a separate build from scratch.

A Dataverse trigger. It fires on row create/update/delete events, unlike a recurrence trigger, which fires on a fixed time schedule regardless of data changes.

An HTTP (request) trigger. It exposes a callable endpoint that an outside system invokes to start the flow, as opposed to a manual trigger that a user starts from within an app.

An approval action. It pauses the flow until a designated approver responds, then routes the flow down an approve or reject path based on that response.

A condition. It evaluates an expression and routes the flow down different sets of actions depending on whether the expression is true or false.

Apply to each. It loops the actions inside it over every item in the specified array or collection.

Set 'Run after' on that action to trigger on the failure, timeout, or skipped conditions of the earlier step. This is the core error-path control mechanism in cloud flows.

A Scope. It wraps multiple actions together so 'Run after' error handling can be applied to the group collectively, instead of configuring it action by action.

A custom prompt. It is a reusable instruction set with defined inputs, distinct from an AI Builder model, which is a Microsoft-trained structured-prediction model rather than an editable instruction.

A prompt template. It is a prebuilt starter structure a maker customizes, saving the effort of writing a fully blank prompt.

Add a knowledge source to the prompt. It supplies grounding data, such as Dataverse tables or documents, that the prompt references when generating its response.

A prompt input. It is a defined variable slot in the prompt that the caller fills in each time the prompt is invoked.

Consume the prompt in the app. It is invoked like a function call from Power Fx, passing whatever prompt inputs the prompt requires.

Consume the prompt in the flow. Add it as an action inside the cloud flow, feeding it inputs gathered from earlier steps in that flow.

A prebuilt AI Builder model, such as invoice processing. It is a Microsoft-trained structured-prediction model, unlike a custom AI Hub prompt, which needs no training but gives less structurally consistent output.

AI Hub prompts are flexible, GPT-style instructions that require no training but produce more variable output. AI Builder models are trained, consistent, and purpose-built for specific structured tasks like document extraction or classification.

A business rule. It is no-code, field-level logic (show, hide, require, validate) that can run on a form or be scoped to the entire table.

An entity-scoped rule applies everywhere the table is used, including server-side validation and every form, while a form-scoped rule only affects the client-side behavior of that one form.

A business process flow (BPF). It provides stage-based guidance across a record's lifecycle, which is different from a business rule that is field-level rather than stage-based.

A formula column. It supports Power Fx-style expressions and evaluation, extending beyond the simpler expression model used by a calculated column.

A calculated column evaluates instantly on read using data from the same row. A rollup column aggregates values from related child rows and recalculates on an asynchronous schedule rather than instantly.

Business required is enforced through business rules or process logic at the UI level and can be bypassed by some advanced tools or APIs. System required is enforced at the platform/database level, and a record cannot be saved at all without that value.

Match the logic's shape to the right tool: field-level validation or defaults call for a business rule, multi-stage record guidance calls for a BPF, and multi-step automation or cross-system integration calls for a cloud flow.