I85PREP
⌘ K
OCI Digital Assistant Professional study materials
OCI Digital Assistant Professional study resources

OCI Digital Assistant Professional practice questions

One question at a time. Build your OCI Digital Assistant Professional understanding.

First 50 of 200 questions. Each one lists the choices, the correct answer, and an explanation. The checker at the top of this page saves progress in this browser when it is available.

Continue in the practice tool above to load more questions, 50 at a time.

Question 1Digital Assistant Design and Building

In Oracle Digital Assistant, which component of a dialog flow defines a reusable unit of conversation logic that can be invoked from multiple places and returns control to its caller?

  1. A channel
  2. A subflow Correct answer.
  3. A dialog flow state
  4. An entity

Explanation

A subflow is a reusable dialog flow that another flow invokes and that returns control to the invoking flow when it completes. States are the individual steps inside a flow, entities extract values, and channels are the messaging interfaces, so none of those provides reusable callable logic.

Question 2Digital Assistant Design and Building

A developer wants a skill to remember a customer's account number across multiple conversation turns within the same session. Which mechanism should be used?

  1. A custom component
  2. A context variable Correct answer.
  3. A channel-specific parameter
  4. A system entity

Explanation

Context variables persist for the duration of a session and are the standard way to carry values such as an account number between states and flows. Entities only extract values at a point in the dialog, channel parameters configure the channel, and custom components perform processing rather than storage.

Question 3Digital Assistant Design and Building

Which statement best describes the role of an entity in Oracle Digital Assistant?

  1. It defines the set of values and validation rules used to extract meaning from user input Correct answer.
  2. It routes the conversation to the correct channel based on the user's device
  3. It authenticates users before the conversation begins
  4. It stores the skill's compiled dialog flow for deployment

Explanation

An entity declares the type of information to extract, such as a date or a list of pizza sizes, along with how values are recognized and validated. Routing, deployment packaging, and authentication are handled by channels, the skill build process, and authentication settings respectively.

Question 4Digital Assistant Design and Building

What is the primary purpose of the System.Intent component in a dialog flow?

  1. To reset the conversation context to its initial state
  2. To validate that a required entity value is present
  3. To resolve the user's utterance to the most likely intent and transition accordingly Correct answer.
  4. To send a greeting message to the user

Explanation

System.Intent performs intent resolution, comparing the user's input against the skill's training utterances and returning the matched intent so the flow can branch. It does not send messages, validate entity values, or reset context.

Question 5Digital Assistant Design and Building

Which artifact in Oracle Digital Assistant groups related intents, entities, and flows into a deployable conversational unit?

  1. A channel
  2. A skill Correct answer.
  3. A component service
  4. A digital assistant

Explanation

A skill is the deployable unit that contains the intents, entities, and dialog flows implementing a specific conversational capability. A digital assistant orchestrates one or more skills, channels expose skills to users, and component services host custom code.

Question 6Digital Assistant Design and Building

A developer needs to ask the user for a date and ensure the value is valid before continuing. Which pair of components is most appropriate?

  1. An entity resolution component such as System.ResolveEntities using a date entity Correct answer.
  2. System.CommonResponse followed by System.Reset
  3. System.Intent followed by System.Text
  4. System.ResolveEntities followed by System.Output

Explanation

Entity resolution components such as System.ResolveEntities prompt for and validate entity values, including date entities, before the flow continues. The other options either handle intent matching, generic messaging, or context reset and do not perform typed entity validation.

Question 7Digital Assistant Design and Building

What does training a skill's intent model accomplish?

  1. It provisions the channel connection to the messaging platform
  2. It compiles the dialog flow into a deployable artifact
  3. It encrypts sensitive data stored in context variables
  4. It teaches the model which utterances map to which intents so it can classify future input Correct answer.

Explanation

Training the intent model uses the example utterances attached to each intent so the classifier can generalize to new user input. Compilation, channel provisioning, and data protection are separate concerns handled elsewhere in the platform.

Question 8Digital Assistant Design and Building

Which option describes a custom component in Oracle Digital Assistant?

  1. A JavaScript snippet embedded directly in a dialog state
  2. A configuration file that maps channels to skills
  3. A reusable code package, typically deployed to a component service, that the dialog invokes Correct answer.
  4. A predefined system entity for extracting numbers

Explanation

Custom components are packaged code, commonly Node.js deployed to a component service, that the dialog invokes to perform logic beyond built-in components. They are not inline dialog snippets, entities, or channel mappings.

Question 9Digital Assistant Design and Building

A dialog flow must branch based on whether the user confirmed a purchase. Which approach correctly evaluates the stored confirmation value?

  1. Add a channel and route the confirmation to it
  2. Add a System.Intent component and check its confidence score
  3. Configure a state transition with a condition that tests the confirmation context variable Correct answer.
  4. Create a new entity and re-prompt the user

Explanation

Conditional transitions evaluate expressions against context variables, so testing the stored confirmation variable in a transition condition is the correct branching mechanism. Intent confidence, entity re-prompting, and channel routing do not evaluate an already captured value.

Question 10Digital Assistant Design and Building

A developer wants the skill to ask a follow-up question only when the user's order total exceeds a threshold stored in a variable. What is the most appropriate design?

  1. Put the condition inside the entity's validation rule
  2. Configure the channel to send the follow-up message
  3. Rely on the intent classifier to select the follow-up intent automatically
  4. Define the condition on the transition into the follow-up state Correct answer.

Explanation

Transition conditions control whether a state is entered, so placing the threshold comparison on the transition into the follow-up state runs the question only when the condition holds. Entity validation, intent classification, and channel configuration do not implement value-based branching.

Question 11Digital Assistant Design and Building

When multiple intents could match a user's utterance, what does the intent resolution result provide to help the dialog decide?

  1. Only the single best intent with no alternatives
  2. The entity values extracted from the previous turn
  3. A ranked list of candidate intents with confidence scores Correct answer.
  4. The raw text of the utterance without classification

Explanation

Intent resolution returns candidate intents ranked with confidence scores so the dialog can apply thresholds or disambiguation logic. It does not return only one intent, raw unclassified text, or entity values from prior turns.

Question 12Digital Assistant Design and Building

Which practice most improves the reliability of an intent classifier for a skill?

  1. Providing a balanced, varied set of example utterances per intent, including realistic phrasing Correct answer.
  2. Increasing the number of channels connected to the skill
  3. Adding many utterances that are near-duplicates of a single phrase
  4. Removing all utterances from intents with low usage

Explanation

Classifier quality depends on varied, realistic, and balanced training utterances that represent how users actually phrase each intent. Near-duplicates add little diversity, deleting utterances removes training signal, and adding channels does not affect classification.

Question 13Digital Assistant Design and Building

A developer needs to collect several related values, such as street, city, and postal code, in one interaction. Which design is most efficient?

  1. Create three separate states each using System.Intent
  2. Hard-code the values in a message component
  3. Ask the user to type all values into a single free-text field and store it unvalidated
  4. Define a composite bag entity and resolve it with an entity resolution component Correct answer.

Explanation

A composite bag entity groups related entities so an entity resolution component can collect and validate them together, often in a single prompt cycle. Separate intent states, hard-coded messages, or unvalidated free text do not provide structured multi-value collection.

Question 14Digital Assistant Design and Building

What is the effect of setting an entity resolution component to allow multiple values for a list entity, such as pizza toppings?

  1. The user must provide exactly one topping per turn
  2. The component can capture several matching values from a single utterance Correct answer.
  3. The dialog automatically creates a new intent for each topping
  4. The entity is ignored if more than one value is mentioned

Explanation

Enabling multi-value resolution lets the component extract multiple valid entity values from one user utterance, such as two toppings mentioned together. It does not force one value per turn, discard extra values, or generate new intents.

Question 15Digital Assistant Design and Building

A developer wants a custom component to call an external REST service and return the result to the dialog. Which statement is correct?

  1. The external service must be registered as an entity before it can be called
  2. The result must be written to a channel configuration parameter
  3. The component can perform the call and return results through its output properties to the dialog Correct answer.
  4. Custom components cannot make network calls and must rely on built-in components

Explanation

Custom components execute code, including HTTP calls to external services, and expose results to the dialog through their defined output properties. They are not restricted from network access, services are not registered as entities, and results are not stored in channel configuration.

Question 16Digital Assistant Design and Building

Which component is best suited to present a set of clickable choices to the user in a channel that supports buttons?

  1. System.Intent
  2. System.CommonResponse Correct answer.
  3. System.Reset
  4. System.Text

Explanation

System.CommonResponse supports rich response types such as choices, cards, and buttons that adapt to the channel's capabilities. System.Text sends plain text, System.Reset clears context, and System.Intent performs intent resolution.

Question 17Digital Assistant Design and Building

A skill must gracefully handle input that matches no intent above the confidence threshold. Which design supports this?

  1. Route all unmatched input to a channel
  2. Remove the confidence threshold so every input matches something
  3. Configure an unresolved intent handler that prompts the user or offers help Correct answer.
  4. Disable intent training until matches improve

Explanation

An unresolved intent handler catches input that fails to meet the confidence threshold so the skill can re-prompt, offer help, or escalate. Removing thresholds causes incorrect matches, channels do not handle unmatched intent logic, and disabling training degrades classification.

Question 18Digital Assistant Design and Building

What is the purpose of the digital assistant layer relative to individual skills?

  1. It compiles custom components for deployment
  2. It stores the training utterances for each skill's intents
  3. It replaces skills by containing all dialog flows directly
  4. It routes user requests to the most appropriate skill and can provide cross-skill behavior Correct answer.

Explanation

A digital assistant sits above skills, routing requests to the best-matching skill and enabling shared behavior across skills. It does not contain the skill flows itself, hold skill training utterances, or compile custom components.

Question 19Digital Assistant Design and Building

A developer needs to reuse the same confirmation logic in three different flows within a skill. What is the recommended approach?

  1. Create a subflow containing the logic and invoke it from each flow Correct answer.
  2. Define a new intent for each flow that needs confirmation
  3. Copy the states into each flow to avoid dependencies
  4. Add the logic to the channel configuration

Explanation

A subflow encapsulates the confirmation logic once and can be invoked from multiple flows, reducing duplication and maintenance risk. Copying states duplicates logic, channel configuration does not hold dialog logic, and new intents do not provide reusable flow logic.

Question 20Digital Assistant Design and Building

Which approach correctly validates that a user-supplied quantity is a positive integer before the order proceeds?

  1. Increase the intent confidence threshold
  2. Apply a validation rule on the quantity entity that rejects non-positive values Correct answer.
  3. Store the value in a channel parameter
  4. Send a confirmation message and assume the value is valid

Explanation

Validation rules on the entity enforce constraints such as positivity during entity resolution, preventing invalid values from advancing the flow. Assumptions, confidence thresholds, and channel parameters do not validate numeric constraints.

Question 21Digital Assistant Design and Building

A dialog flow needs to prompt the user differently depending on whether the channel supports rich cards. What is the most appropriate mechanism?

  1. Create a separate skill for each channel
  2. Inspect channel type information available to the dialog and branch accordingly Correct answer.
  3. Hard-code card output for every channel
  4. Rely on System.Intent to detect the channel

Explanation

The dialog can inspect channel-related information exposed in the runtime context and branch so that rich responses are used only where supported. Separate skills per channel, intent resolution, or hard-coded card output do not adapt output to channel capabilities.

Question 22Digital Assistant Design and Building

A developer observes that a skill frequently misroutes utterances between two similar intents. Which analysis step is most appropriate first?

  1. Add more channels to increase training data
  2. Lower all confidence thresholds to zero
  3. Delete one of the two intents permanently
  4. Review the example utterances for overlap and add disambiguating examples Correct answer.

Explanation

Similar intents often share overlapping utterances, so reviewing and refining the training examples to distinguish them is the correct first diagnostic step. Adding channels does not add training data, deleting an intent may remove needed capability, and zero thresholds worsen misrouting.

Question 23Digital Assistant Design and Building

A composite bag entity must skip prompting for a value already supplied earlier in the conversation. Which design achieves this?

  1. Set the intent confidence threshold to one
  2. Remove the field from the bag entity entirely
  3. Pre-populate the bag entity's fields from existing context variables before resolution Correct answer.
  4. Re-prompt for every field regardless of prior input

Explanation

Pre-populating the bag's fields with values already captured lets the entity resolution component skip prompting for those fields while still collecting the rest. Re-prompting ignores prior input, removing the field loses the requirement, and the confidence threshold is unrelated.

Question 24Digital Assistant Design and Building

A custom component returns an error that the dialog must handle without crashing the conversation. Which design is most robust?

  1. Ignore the error and continue as if the call succeeded
  2. Define output properties for status and message, then branch on them in the dialog to handle failure Correct answer.
  3. Throw an unhandled exception so the channel restarts
  4. Store the error in a channel parameter for later review

Explanation

Returning structured status and message outputs lets the dialog branch on the outcome and present a graceful recovery path. Ignoring errors hides failures, unhandled exceptions disrupt the conversation, and channel parameters are not the right place for runtime error handling.

Question 25Digital Assistant Design and Building

A developer wants a skill to ask for clarification when the top two intent confidence scores are close. Which capability enables this?

  1. Removing all but one intent from the skill
  2. Setting every intent's confidence to a fixed value
  3. Disabling intent resolution entirely
  4. Using the ranked candidate intents from intent resolution to trigger a disambiguation prompt Correct answer.

Explanation

Intent resolution returns ranked candidates, so the dialog can compare the top scores and, when they are close, ask the user to disambiguate. Disabling resolution, fixing confidence values, or removing intents eliminates the information needed to detect ambiguity.

Question 26Digital Assistant Design and Building

In Oracle Digital Assistant, which artifact holds the training utterances that teach a skill to recognize what a user wants?

  1. The intent definition within the skill's natural language model Correct answer.
  2. The dialog flow's state transition table
  3. The digital assistant's routing rule set
  4. The channel's message transformation template

Explanation

Intents are defined in the skill's natural language model, and each intent owns the example utterances used to train the classifier. The dialog flow consumes intent results but does not store training data, and channel or routing configuration does not affect intent recognition.

Question 27Digital Assistant Design and Building

A developer wants to define a structured value such as a shipping address with named subfields inside a skill. Which modeling construct is intended for this?

  1. A static list entity
  2. A system entity such as DATE
  3. An Apache FreeMarker template
  4. A composite bag entity Correct answer.

Explanation

A composite bag entity groups related entity items so the skill can collect and validate a structured value such as an address. List entities hold flat enumerations, system entities cover built-in types like dates, and FreeMarker templates format output rather than model data.

Question 28Digital Assistant Design and Building

Which statement describes how a skill communicates its result back to the digital assistant that invoked it?

  1. It publishes a message to the channel's webhook endpoint
  2. It restarts the digital assistant's main flow from the beginning
  3. It writes a row directly into the digital assistant's routing table
  4. It sets a transition action that returns control to the digital assistant with an outcome Correct answer.

Explanation

A skill returns control to its caller through a transition action such as return, which carries an outcome the digital assistant can act on. Skills do not write routing configuration, call channel webhooks, or restart the assistant's flow to report completion.

Question 29Digital Assistant Design and Building

Which built-in system entity should a developer reference to capture a monetary amount like "twenty-five dollars" from a user utterance?

  1. PERSON
  2. DATE
  3. CURRENCY Correct answer.
  4. LOCATION

Explanation

The CURRENCY system entity recognizes monetary amounts and their units, so it is the correct choice for capturing a price. DATE, PERSON, and LOCATION each recognize unrelated semantic types.

Question 30Digital Assistant Design and Building

What is the role of the dialog flow's context variables when a skill runs?

  1. They select which channel renders the response
  2. They define the training utterances for each intent
  3. They configure the skill's connection to an external database
  4. They store values exchanged between states during the conversation Correct answer.

Explanation

Context variables hold data such as user input, entity values, and intermediate results so that states can read and pass information along the conversation. They are not training data, channel selectors, or database connection settings.

Question 31Digital Assistant Design and Building

A developer needs to publish a skill so users can reach it through a digital assistant. Which sequence reflects the required steps?

  1. Attach the skill to a channel first, then define its intents
  2. Configure the skill, train it, then attach it to a digital assistant and publish the assistant Correct answer.
  3. Create the digital assistant first, then define the skill's dialog flow inside the assistant
  4. Publish the skill before training it, then attach it to a channel

Explanation

A skill must be configured and trained before it can be attached to a digital assistant, and publishing the assistant makes the combined experience available. Training after publishing or channel attachment before intent definition inverts the dependency order.

Question 32Digital Assistant Design and Building

Which component type should a developer use to send a plain text prompt asking the user for their order number?

  1. A switch component with a condition
  2. A send component with a text message Correct answer.
  3. A reset component targeting the context
  4. A resolve component with a list entity

Explanation

The send component outputs a message to the user, making it the direct way to present a text prompt. Resolve components collect entity values, switch components branch on conditions, and reset components clear state.

Question 33Digital Assistant Design and Building

Which feature lets a skill recognize different ways of expressing the same request without creating a separate intent for each phrasing?

  1. Lowering the confidence threshold to zero
  2. Disabling entity resolution
  3. Adding multiple example utterances to a single intent Correct answer.
  4. Creating a separate intent for every phrasing

Explanation

An intent's classifier learns from all of its example utterances, so varied phrasings can be grouped under one intent. Splitting phrasings into many intents fragments the model, and threshold or entity settings do not add linguistic coverage.

Question 34Digital Assistant Design and Building

A developer wants to branch on the value of an entity captured earlier in the same flow. Which component is designed for evaluating conditions and choosing the next state?

  1. The send component
  2. The service component
  3. The text component
  4. The switch component Correct answer.

Explanation

The switch component evaluates one or more conditions against context variables and routes the conversation to the matching transition. Send and text components output messages, and service components invoke backend integrations.

Question 35Digital Assistant Design and Building

Which approach correctly stores a value returned by a custom component so later states in the same dialog can use it?

  1. Return the value in the component's response and map it to a context variable Correct answer.
  2. Set the value as the skill's display name
  3. Write the value into the intent's training utterances
  4. Emit the value as a channel-specific message template

Explanation

Custom components communicate results through their response payload, and the dialog maps that payload to context variables for downstream states. Training data, skill metadata, and channel templates are not runtime storage for component results.

Question 36Digital Assistant Design and Building

A developer wants the skill to ask for a missing entity value only when the user has not already provided it. Which mechanism handles this automatically?

  1. An intent reclassification step
  2. A hard-coded send component that always asks
  3. A switch component with a default transition
  4. Entity resolution prompting within a resolve component Correct answer.

Explanation

The resolve component prompts only for entity values that are still unresolved, using the entity's prompt definitions. A hard-coded send would ask even when the value exists, switch components do not collect entity values, and reclassification does not fill missing slots.

Question 37Digital Assistant Design and Building

Which design keeps a skill's dialog flow maintainable when the same greeting message must appear at several entry points?

  1. Store the greeting in the channel's webhook configuration
  2. Define the greeting once in a reusable state and transition to it from each entry point Correct answer.
  3. Hard-code the greeting inside a custom component
  4. Copy the greeting text into every entry state

Explanation

A reusable state referenced from multiple entry points centralizes the message so it changes in one place. Duplicating text across states, embedding it in code, or placing it in channel configuration all scatter the logic and complicate maintenance.

Question 38Digital Assistant Design and Building

A developer wants to display a list of choices as buttons on channels that support them while still working on text-only channels. Which approach is most appropriate?

  1. Send a plain text message listing the choices manually
  2. Use a switch component with no user prompt
  3. Rely on the user to type a free-form answer with no options shown
  4. Send a message with choice options so the channel renders buttons when supported and text otherwise Correct answer.

Explanation

Choice-type messages let each channel render the best available representation, such as buttons or a numbered text list. Manual text, silent switching, or free-form input all sacrifice the channel-aware rendering the requirement asks for.

Question 39Digital Assistant Design and Building

A developer needs to invoke an external REST endpoint from a dialog flow without writing custom code. Which component supports this directly?

  1. The service component configured with a REST integration Correct answer.
  2. The resolve component with a list entity
  3. The switch component with an expression
  4. The send component with a text message

Explanation

The service component invokes configured integrations, including REST services, and exposes the response to the flow. Send, switch, and resolve components handle messaging, branching, and entity collection respectively.

Question 40Digital Assistant Design and Building

What is the effect of defining an entity's prompt as a message with a range of retry prompts?

  1. The skill blocks the channel from rendering buttons
  2. The skill can rephrase its request when the user's answer cannot be resolved Correct answer.
  3. The skill lowers the intent confidence threshold automatically
  4. The skill skips the entity entirely after the first attempt

Explanation

Retry prompts let the skill ask again in different wording when resolution fails, improving the chance of a valid answer. They do not skip the entity, alter intent thresholds, or disable channel rendering.

Question 41Digital Assistant Design and Building

A developer wants to log diagnostic details when a custom component fails so the issue can be investigated later. Which practice is most appropriate?

  1. Suppress the error and continue silently
  2. Store the details in the intent's training utterances
  3. Emit the diagnostic details to the end user as a chat message
  4. Write structured entries to the component's logger with context identifiers Correct answer.

Explanation

Structured logging with identifiers lets developers trace failures without exposing internals to users. Surfacing diagnostics in chat leaks implementation details, polluting training data is meaningless, and silent suppression hides defects.

Question 42Digital Assistant Design and Building

Which configuration determines whether a digital assistant hands a user's message to a particular skill?

  1. The custom component's compiled artifact size
  2. The skill's dialog flow variable names
  3. The digital assistant's skill routing rules and their confidence thresholds Correct answer.
  4. The channel's message character limit

Explanation

Routing rules in the digital assistant decide which skill receives a message, often using confidence thresholds to pick among candidates. Variable names, channel limits, and artifact size do not influence routing decisions.

Question 43Digital Assistant Design and Building

A developer wants the skill to confirm a destructive action before proceeding. Which flow pattern implements this correctly?

  1. Resolve a yes/no entity, then use a switch on its value to proceed or cancel Correct answer.
  2. Delete the context variable before asking
  3. Rely on the intent classifier to cancel automatically
  4. Send the confirmation text and proceed regardless of the answer

Explanation

Confirmation requires capturing the user's yes/no answer and branching on it so the action runs only when confirmed. Proceeding regardless ignores the answer, classifiers do not enforce confirmation, and clearing context destroys the data needed to act.

Question 44Digital Assistant Design and Building

A skill must handle an unexpected exception thrown during entity resolution so the conversation continues gracefully. Which design is most robust?

  1. Define an error transition handler that sends a fallback message and offers recovery Correct answer.
  2. Restart the entire digital assistant for every error
  3. Ignore the exception and leave the user waiting
  4. Let the exception terminate the session immediately

Explanation

Error transition handlers catch failures and let the skill respond with a fallback and a way forward, preserving the conversation. Terminating, hanging, or restarting the assistant all degrade the user experience unnecessarily.

Question 45Digital Assistant Design and Building

A developer wants to test a skill's dialog flow in isolation before wiring it into the digital assistant. Which capability supports this?

  1. The channel's end-user device simulator
  2. The digital assistant's production channel endpoint
  3. The skill's own test/preview environment for the dialog flow Correct answer.
  4. The custom component's unit test harness only

Explanation

Skills provide a test or preview environment that exercises the dialog flow and model without requiring the digital assistant. Production endpoints, component-only harnesses, and device simulators do not isolate the skill's flow.

Question 46Digital Assistant Design and Building

What is the purpose of the digital assistant's main flow relative to its skills?

  1. It replaces each skill's dialog flow entirely
  2. It handles cross-skill concerns such as greetings, routing, and unresolved input Correct answer.
  3. It defines the entity types shared by all skills
  4. It stores the training utterances for every attached skill

Explanation

The digital assistant's main flow manages conversation-level behavior such as greetings, skill selection, and fallback when no skill handles the input. It does not host skill training data, replace skill flows, or define their entities.

Question 47Digital Assistant Design and Building

A skill's entity resolution keeps failing because users supply values in an unsupported format. Which analysis step best identifies the gap?

  1. Raise the intent confidence threshold to filter the input
  2. Remove the entity from the flow entirely
  3. Shorten the entity's prompt text
  4. Review the failed resolution utterances and extend the entity's value definitions or custom resolver Correct answer.

Explanation

Failed resolution logs reveal the formats the entity could not interpret, guiding additions to value lists or a custom resolver. Threshold changes affect intent selection, removing the entity abandons the requirement, and shorter prompts do not improve parsing.

Question 48Digital Assistant Design and Building

A developer wants the same skill to behave differently for two user segments without duplicating the entire flow. Which design best achieves this?

  1. Hard-code segment checks inside every send message
  2. Use separate digital assistants with no shared logic
  3. Clone the skill once per segment and maintain both copies
  4. Branch within shared states on a segment context variable set at the start of the session Correct answer.

Explanation

A single segment variable evaluated at key branch points keeps one flow authoritative while varying behavior. Cloning skills, scattering checks, or splitting assistants multiplies maintenance and risks divergence.

Question 49Digital Assistant Design and Building

When a user's utterance matches an intent but the required entity value is absent, what is the most reliable way for the dialog to continue?

  1. Treat the missing value as an error and end the session
  2. Resolve the missing entity value with a targeted prompt before proceeding Correct answer.
  3. Reclassify the intent until a value appears
  4. Assume a default value without informing the user

Explanation

Entity resolution prompts specifically for the missing value, keeping the conversation on track. Reclassification does not create values, ending the session is unnecessarily harsh, and silent defaults can cause incorrect actions.

Question 50Digital Assistant Design and Building

A developer wants to compare two candidate phrasings of a prompt to see which yields more successful entity resolution. Which approach is most sound?

  1. Rely on the default prompt with no evaluation
  2. Choose the longer prompt because length guarantees clarity
  3. Run both prompt variants against representative utterances and compare resolution success rates Correct answer.
  4. Ship both prompts simultaneously in the same state

Explanation

Measuring resolution success across representative inputs gives evidence for choosing between prompt variants. Length is not a quality proxy, sending both prompts at once confuses users, and accepting defaults skips evaluation entirely.