The AB-410 exam certifies your ability to design, build, and implement AI-powered business solutions in Microsoft Power Platform, combining Copilot, natural language prompts, and low-code tools to create Dataverse data models, model-driven apps, canvas apps, and cloud flows. Currently in beta, it targets app builders who already work with Power Apps, Power Automate, and Dataverse and now need to weave in built-in agents, AI Hub prompts and models, and Copilot-generated pages rather than build every screen and flow by hand. It tests three areas: Create a foundation for intelligent applications (25-30%), Create intelligent applications (25-30%), and Build business application logic and automation (40-45%), with the heaviest weight on cloud flows, AI Hub prompts and models, and declarative business logic like business rules and business process flows. A passing score is 700 out of 1000 across a 120-minute exam. Many questions reward Microsoft's preferred pattern for a scenario: reach for a built-in agent or an AI Hub prompt before hand-coding logic, and pick the lightest mechanism, a business rule over a business process flow over a full cloud flow, that still meets the requirement.
What This Cheat Sheet Covers
This topic spans 25 focused tables and 222 indexed concepts, 220 flashcards, 8 practice tests with 294 questions. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
A jump-to index of every table row in this cheat sheet.
An interactive map of every table and concept in this topic.
Table 1: Design AI-Enabled Power Platform Solutions
AB-410's Create a Foundation for Intelligent Applications area, task Design Microsoft Power Platform solutions by using AI-enabled tools: matching a business requirement to the right AI-enabled component, evaluating Microsoft's built-in agents before building one from scratch, and knowing the extensibility options once the built-in and low-code toolset falls short.
| Concept | Example | Description | |
|---|---|---|---|
Invoice processing model reads vendor name, total, and due date from a scanned PDF with zero training data | Ready to use immediately for common scenarios (invoices, receipts, ID documents, sentiment, key phrases), no training data needed. β’Available in Power Apps and Power Automate. β’Try this before a custom model whenever the scenario is a common one. | ||
Document processing model trained on 50 sample warranty claim forms to extract claim number and part list | Trained on your own proprietary data (documents, text, images) for a need no prebuilt model covers. β’Includes document processing, entity extraction, prediction, object detection, and bring-your-own Azure Machine Learning models. β’Not to be confused with a prebuilt model: it needs training data before it can be used. | ||
A support agent that must run in Teams, web chat, and Microsoft 365 Copilot is built as a Custom agent, not a Declarative agent | Custom agents are standalone, built and published from Copilot Studio to Teams, web chat, and Microsoft 365 Copilot. β’Declarative agents only extend Microsoft 365 Copilot itself, built with Agent Builder or the Microsoft 365 Agents Toolkit, and deployed as a Teams app. β’Not to be confused with each other: a declarative agent cannot run as a standalone, multi-channel experience. | ||
Install the My Company Policy template from the Agent Library to answer HR and IT questions in minutes | Marketplace-style catalog of complete, production-ready agent templates for Copilot Studio and Microsoft 365 Copilot. β’Deploy a template instead of building from scratch to move deployment from weeks to minutes. β’Fully customizable after deployment (instructions, knowledge sources, branding). | ||
Import the Document Extraction component into an existing custom agent instead of deploying a whole new template | Modular building blocks (Research, Document Extraction, Content Synthesis, and more) imported into a maker's own agent and called from a Topic. β’Not a complete agent by itself; not to be confused with a full Agent Library template. | ||
Deploy the After Hours Service Agent to give customers round-the-clock coverage without new overnight staff | First-party agents built on Dynamics 365 Customer Service and Contact Center capabilities plus Copilot Studio. β’Evaluate these before building a similar custom Copilot Studio agent for a routine service scenario (case categorization, after-hours coverage, follow-ups). | ||
Wrap an internal shipping vendor's REST API with an OpenAPI definition to create a reusable connector | A wrapper around a REST or SOAP API, described with an OpenAPI definition or Postman collection, for a service with no prebuilt connector. β’A connector built in Power Apps or Power Automate is reusable across Copilot Studio; one built in Logic Apps is not automatically portable. β’Must be certified before it can be shared with all users. | ||
Add the Dynamics 365 connector to an agent so it can look up a customer's open cases before answering | The mechanism Copilot Studio agents use to ground responses and take actions against Microsoft 365, Dynamics 365, Microsoft Fabric, and non-Microsoft data. β’Prefer an existing prebuilt connector; reach for a custom connector only when no prebuilt one exists. | ||
A synchronous plug-in on Account update blocks the save until it validates the credit limit | Custom .NET event handlers registered on a Dataverse data event (create, update, delete), run synchronously or asynchronously. β’Consider a declarative option first (calculated/rollup fields, custom actions, Power Automate); use a plug-in only when a declarative process cannot meet the requirement. β’A synchronous plug-in blocks the operation until it finishes, so keep it fast. | ||
Replace a numeric text column on a form with a slider control code component | Code components (HTML/CSS/TypeScript) that replace or enhance a control on a model-driven or canvas app screen, unlike static HTML web resources. β’A component that talks directly to an external service, not through a connector, makes the whole app premium, so end users need full Power Apps licenses. | ||
LibraryGenre( SelectedGenre: Text ): LibraryType = Filter( Library, Genre = SelectedGenre ) | Parameterized, reusable Power Fx formulas that extend named formulas and behave like built-in functions. β’The low-code option to try before writing a Dataverse plug-in or Azure Function when the logic can be expressed as a formula. β’Wrap the body in curly braces only when the function needs a behavior side effect like Set. |