Building & Automating Flows with AI

What description pattern does Copilot in Power Automate recommend when a maker types a request to create a new cloud flow?

• A"While X is true, do Y", describing a continuous condition and a looping action
• B"If X, then never Y", describing an event to avoid and an action to block
• C"When X happens, do Y", describing the triggering event and the resulting actions✓
• D"After X finishes, undo Y", describing a completed action and a reversal step

Answer: Copilot in Power Automate works best with descriptions that follow the pattern 'When X happens, do Y', being specific about which services are involved and what data should flow between them, so it can propose an appropriate trigger and action sequence. The other patterns describe negation, continuous looping or reversal, none of which match the trigger-then-action structure Copilot is designed to interpret. Once the maker selects Generate, Copilot processes the description and proposes a flow structure for review.

After Copilot proposes a flow structure from a natural language description in Power Automate, what should the maker do next?

• APublish the proposal directly to production without reviewing it, since Copilot validates it automatically
• BAccept the proposal immediately, because Copilot cannot be asked to make further changes
• CReview the proposal, confirming the trigger, action order and branching logic before accepting it✓
• DDiscard the proposal and rebuild the flow manually, because Copilot only works for the first draft

Answer: Copilot displays a proposed flow structure showing the trigger, the sequence of actions, and any detected conditions or branching, and the maker should review this carefully to confirm the trigger matches the described event, the action order is correct, and the branching logic reflects their intent. If something is wrong, the maker can continue the conversation with follow-up prompts, and Copilot updates the proposal accordingly rather than being limited to one attempt. Only once the maker is satisfied is the flow accepted into the designer for final review.

A Contoso employee wants to build a desktop flow that opens Excel, reads pricing data, and updates a pricing portal website, and prefers to describe the process in words rather than record it. Which Power Automate for Desktop feature should they use?

• ARecord with Copilot, narrating the steps while the screen is shared
• BCopilot in Power Automate for Desktop, describing the process in the Copilot Chat panel✓
• CThe Recurrence trigger, scheduling the flow to run at a fixed time each day
• DThe PowerApps trigger, passing the pricing data from a canvas app

Answer: Copilot in Power Automate for Desktop, accessed through the Copilot Chat panel, can generate a sequence of desktop flow actions directly from a natural language description of the process, which is what this employee prefers. Record with Copilot (the AI Recorder) instead builds the flow by capturing a screen-shared demonstration with voice narration, which suits makers who find it easier to show the process than describe it. Recurrence and the PowerApps trigger are cloud flow trigger types and do not build the action sequence itself.

A maker finds it easier to demonstrate a legacy-application process than to describe it in text. Which Power Automate for Desktop feature is designed for this situation?

• AThe Copilot panel in the cloud flow designer at make.powerautomate.com
• BThe flow checker, which validates a flow's configuration before saving
• CRecord with Copilot, also called the AI Recorder✓
• DCopilot in Power Automate for Desktop, using only typed descriptions

Answer: Record with Copilot, also called the AI Recorder, lets a maker build a desktop flow by sharing their screen and narrating the steps while performing them, which the recorder then converts into named desktop flow actions. This suits legacy applications where it's difficult to describe the exact UI elements in text but easy to demonstrate visually, unlike the typed-description Copilot experience. The cloud flow designer's Copilot panel and the flow checker both relate to cloud flows rather than to building desktop flows from a demonstration.

Which requirement applies specifically to Record with Copilot (the AI Recorder) in Power Automate for Desktop, but not to describing a desktop flow to Copilot in text?

• AIt requires the flow to already exist, while the text-based Copilot experience can only create new flows
• BIt requires screen-sharing permissions and an active microphone, and currently supports English only✓
• CIt requires a Dataverse connection, while the text-based Copilot experience works only with SharePoint
• DIt requires a Power Automate Premium licence, while the text-based Copilot experience is free for all users

Answer: Record with Copilot is a preview feature that currently supports English and requires screen-sharing permissions and an active microphone, since it captures the maker's voice narration and on-screen actions during a live demonstration. Describing a process in the Copilot Chat panel needs neither of these, since it works from typed text alone. Licensing tiers, needing a pre-existing flow, and a Dataverse connection are not distinctions the unit draws between the two features.

Which two modes does the Power Automate cloud flow designer offer for building a flow?

• AVisual designer mode and Record with Copilot mode
• BVisual designer mode and AI Recorder mode
• CCopilot-assisted mode and Power Automate for Desktop mode
• DVisual designer mode and Copilot-assisted mode✓

Answer: The cloud flow designer at make.powerautomate.com offers visual designer mode, where triggers and actions appear as cards on a canvas, and Copilot-assisted mode, where the Copilot panel sits alongside the visual designer so the maker can keep building through natural language. Record with Copilot and the AI Recorder are Power Automate for Desktop features for building desktop flows from a demonstration, not cloud flow designer modes, and Power Automate for Desktop is a separate desktop application rather than a cloud flow designer mode.

In the anatomy of a cloud flow, what are dynamic content values?

• ACards on the designer canvas that represent the trigger and each action in sequence
• BAuthenticated links to external services established using the maker's credentials
• CFormulas written in the Power Automate expression language that perform calculations on values
• DData outputs from previous steps in the flow that can be referenced in later action configurations✓

Answer: Dynamic content is data output from previous steps in the flow, such as the subject of the triggering email or the ID of a created SharePoint item, and it can be referenced in the configuration of later actions. Expressions are instead formulas in the Power Automate expression language used for calculations and transformations on such values, connections are the authenticated links to external services rather than the data itself, and the cards on the canvas represent the trigger and actions, not the data they output.

A flow must retry an operation repeatedly until a status changes to 'Complete', rather than a fixed number of times. Which loop action fits this requirement?

• ADo until, which repeats a set of actions until a specified condition becomes true✓
• BSwitch, which routes to a different set of actions for each matching case
• CParallel branch, which runs multiple action sequences at the same time
• DApply to each, which iterates through every item in a collection

Answer: Do until repeats a set of actions until a specified condition becomes true, making it suited to polling scenarios where a flow should retry an operation until a status changes or a call succeeds. Apply to each instead loops a fixed number of times, once per item in a known collection, rather than until a condition is met. Switch routes to different actions based on a value rather than looping, and a parallel branch runs independent actions simultaneously rather than repeating one action.

When a new customer is onboarded at Contoso, a flow creates a Dataverse record, adds an Excel row, sends a welcome email, and posts a Teams notification, all starting at the same time rather than one after another. Which building block enables this?

• AParallel branches, which run multiple action sequences concurrently✓
• BA Scope, which groups actions for organisational clarity and error handling
• CA Switch control, which routes to a different case for each value
• DApply to each, which loops through every item in a collection

Answer: Parallel branches let independent actions run concurrently instead of the flow's default sequential order, reducing total execution time when tasks such as creating a record, adding a row, sending an email and posting a notification don't depend on each other's output. Apply to each loops over a collection rather than starting unrelated actions together, a Switch control branches based on a value rather than running actions simultaneously, and a Scope groups actions for organisation and error handling rather than for concurrency.

Before enabling a newly built cloud flow for regular use, a maker wants automatic confirmation that no required fields are missing and no expressions are invalid. Which designer feature provides this?

• AThe flow checker, which validates the flow configuration before saving✓
• BThe Templates gallery, which offers prebuilt starting points for common scenarios
• CRun history, which logs every run with timestamps and step-by-step execution details
• DThe Test button, which runs the flow with either a manual trigger or previous run data

Answer: The flow checker validates a flow's configuration before saving, identifying missing required fields, invalid expressions, or other configuration errors that would cause the flow to fail. The Test button instead actually executes the flow to show each step's real execution status and data, run history is a log of past runs available after the flow has already been used, and the Templates gallery offers prebuilt starting points for new flows rather than validating an existing one.