Domain 1: Describe the Business Value of Microsoft Power Platform
- Microsoft Power Platform is a low-code suite - Power Apps, Power Automate, Power Pages and Microsoft Copilot Studio - built on Microsoft Dataverse and connected to other systems through connectors. Note that Power BI is not part of what this exam measures.
- The core business value is empowering 'citizen developers' (non-professional makers) to build apps, automations, and reports with visual designers and Excel-like formulas instead of traditional code.
- Power Apps delivers business value by letting the people who understand a process build the application for it, in days rather than through a development backlog: a canvas app for a task-shaped mobile process, a model-driven app for a data-shaped one.
- Power Automate delivers value by removing repetitive manual steps: cloud flows connect services that already have APIs, and desktop flows automate older applications that do not, so an organisation can automate a legacy system without replacing it.
- Microsoft Dataverse is the secure, cloud-based relational data platform that stores business data and provides logic, security, and compliance for Power Platform solutions.
- Connectors are pre-built integration points to hundreds of Microsoft and third-party data sources and services; Power Platform is not limited to Microsoft data.
- Power Pages delivers value by extending a Dataverse solution to people outside the organisation - customers, partners, citizens - through a secure external website, so external users can submit and track information without an internal account.
- Generative AI features across Power Platform are now business-value content in their own right: describing an app or a flow in natural language and having it drafted, and having AI suggest tables and columns for a solution, which lowers the barrier further than low-code alone did.
- Copilot Studio delivers value by letting an organisation build agents that answer questions and take action against its own systems and content, so routine requests are handled without a person and without a traditional chatbot development project.
- Power Platform integrates natively with Microsoft 365 (Teams, SharePoint, Outlook, Excel) and Azure, and offers embedded experiences so apps and flows run inside those products.
- Low-code development reduces cost and time-to-value by letting business teams solve departmental problems quickly, easing the backlog on professional development teams.
Domain 2: Manage the Microsoft Power Platform Environment
- Dataverse differs from a traditional database in that it is more than storage: it ships with a security model, business logic, auditing and an API, and it is managed rather than administered. You model tables and relationships instead of provisioning servers, indexes and backups.
- In Dataverse a table (formerly 'entity') stores data as rows (records) and columns (fields/attributes); each column has a defined data type.
- Standard tables (e.g. Account, Contact) ship with Dataverse; custom tables are created to store organization-specific data.
- Each Dataverse environment contains a single database that holds all of that environment's tables and records.
- Relationships associate rows across tables: one-to-many (a customer with many orders), many-to-one, and many-to-many (students and courses, managed via an implicit junction table).
- A lookup column stores a reference to a row in another table, implementing the 'many' side of a one-to-many relationship.
- An Autonumber column auto-generates unique, sequential, human-readable identifiers (with optional prefix/format) on row creation, ideal for ticket or order numbers.
- Dataverse is relational and supports referential integrity, business logic, and security, making it more than a simple flat data store.
- Forms control how a single row is displayed and edited, and views control which rows a list shows and which columns appear. Both are configured rather than coded, and both are what a model-driven app builds its interface from.
- Dataverse offers layered options for business logic: business rules for no-code field behaviour, calculated and rollup columns for derived values, Power Fx formula columns for expression-based logic, and cloud flows or plug-ins when the requirement goes beyond what configuration can do.
- Business rules are no-code logic that set field values, set defaults, show error messages, make fields required, or show/hide fields based on conditions without writing code.
- AI can now create and edit the data model itself: describe the business need in natural language and Dataverse proposes tables, columns and relationships, or attach a file such as a spreadsheet and let it infer the structure. The maker reviews and adjusts what is proposed rather than starting from an empty table.
- Environments are logical containers that isolate apps, flows, and data; organizations use separate Development, Test, and Production environments with their own security, governance, and DLP settings.
- Environment types matter: production for live workloads, sandbox for testing and safe resets, developer for a single maker, trial for time-limited evaluation, and default, which every tenant gets and every user can access - which is exactly why the default environment is the wrong place for governed work.
- The Power Platform security model layers together: Microsoft Entra ID authenticates the user, an environment's access controls decide who gets in, Dataverse security roles grant privileges at access levels from User up to Organization, and business units and teams decide the ownership boundary.
- Data Loss Prevention (DLP) policies classify connectors into Business, Non-business, and Blocked groups; data cannot flow between connectors in different groups within the same app or flow.
- Governance features include environment-level DLP, tenant-wide DLP, sharing limits, and solution checker enforcement to give admins control as adoption scales.
- Power Platform supports data privacy and accessibility obligations directly: customer-managed keys and data residency options for where data lives, auditing for who touched it, and an accessibility checker in the app designers so makers catch contrast, labelling and screen-reader problems before publishing.
- Monitoring and analytics live in the Power Platform admin center: adoption and usage analytics for apps and flows, capacity reporting for storage, and Application Insights integration when telemetry needs to go somewhere an engineering team already watches.
- The Power Platform admin center is where administrators manage environments, DLP policies, data integration, capacity, and analytics.
- The Microsoft 365 admin center (not the Power Platform admin center) is used to manage user accounts, groups, and licenses.
- Solutions are containers that package customizations (apps, flows, tables, columns, plugins) so they can be transported between environments for ALM.
- Unmanaged solutions can be directly edited in the target environment and are used for development; managed solutions cannot be directly customized and are used for test/production deployment.
- Power Platform pipelines bring application lifecycle management in-product: define a pipeline with its stages, and makers deploy solutions from development to test to production from inside the environment, with the deployment recorded, rather than exporting and importing files by hand.
- The Center of Excellence (CoE) Starter Kit is a free Microsoft toolkit of apps, flows, and dashboards that helps organizations monitor, govern, and nurture Power Platform adoption.
- Connectors are the integration layer that abstracts API communication, authentication, and data transformation between apps/flows/dashboards and external services.
- Connector types: standard (included with base licenses), premium (require extra licensing), and custom (built by the organization for REST APIs lacking a connector).
- An on-premises data gateway securely relays data between cloud Power Platform services and on-premises sources behind a firewall, using an encrypted outbound connection without opening inbound ports.
Domain 3: Demonstrate the Capabilities of Power Apps
- Canvas apps give the maker full pixel-level control over screen layout, starting from a blank canvas or template, and can connect to many data sources (Dataverse, SharePoint, Excel/OneDrive, SQL, and more).
- Model-driven apps derive their UI, forms, views, and navigation automatically from the underlying Dataverse data model and require Dataverse as the data source.
- Choose by the shape of the problem: canvas when the layout and the task flow matter and the data may come from anywhere, model-driven when the data model is the application and you want forms, views and navigation generated from it.
- Model-driven apps are built from components including tables, forms, views, charts, dashboards, and business process flows defined in Dataverse.
- Power Apps formulas use Power Fx, a low-code expression language similar in syntax to Excel formulas, making it accessible to spreadsheet users.
- Common Power Fx functions include Filter(), LookUp(), Patch() (create/update records), Navigate() (move between screens), and Collect() (build collections).
- Navigate() moves between canvas-app screens, e.g. Navigate(Screen2, ScreenTransition.Fade), with transitions like Fade, Cover, UnCover, and None.
- A Gallery control displays a scrollable list of multiple records using a repeating template layout.
- A Form control (display/edit) shows, edits, or creates a single record from a data source and supports validation.
- The OnStart property runs when an app loads and is used to set global variables or preload initial data; OnVisible runs each time a screen is shown.
- Canvas apps support offline capabilities and device features like the camera, making them ideal for mobile scenarios such as photographing receipts.
- Canvas apps can be auto-generated from a data source (e.g. a SharePoint list or Dataverse table) to quickly create a three-screen browse/detail/edit app.
- Power Apps provides responsive design and a cross-platform runtime so apps run in a browser and in the Power Apps mobile app on phones and tablets.
- Component libraries let makers create reusable UI components that can be shared across multiple apps for consistency.
- Plan designer starts from the business problem rather than the screen: describe the process and it produces a plan with the user roles, the requirements and the suggested Power Platform components - apps, tables, flows and agents - which you then refine and build out.
- Code apps are for when low-code is not the right fit: a developer builds the application with their own code and framework while still using Power Platform for hosting, connectors, Dataverse and governance, so it remains a managed part of the estate rather than a separate system.
- AI can create a canvas app from a description, an image of a design, or an existing data source, producing screens and controls that the maker then edits. It shortens the blank-canvas problem rather than replacing the maker.
- AI can likewise create a model-driven app by proposing the Dataverse tables and relationships the described process needs and generating the app over them, so the data model and the app arrive together.
- The vibe experience is building an AI-powered app conversationally: you describe what you want in natural language, see it built, and keep refining by describing changes rather than editing properties, with the AI capability built into the result rather than added afterwards.
Domain 4: Demonstrate the Capabilities of Power Automate
- A cloud flow runs in the cloud and connects services and data sources; the three trigger types are automated (event), instant (manual/button), and scheduled (recurrence).
- Automated cloud flows start when a specific event occurs, such as an email arriving in Outlook, a record created in Dataverse, a file added to OneDrive, or an item created in SharePoint.
- Instant (button) flows are triggered manually on demand, including being called from a Power Apps canvas app button or a mobile/Teams button.
- Scheduled flows run on a recurring time-based schedule (e.g. daily at 8:00 AM, hourly, or weekly) using a recurrence trigger.
- Every flow is built from a trigger (the event that starts it) plus one or more actions (the operations it performs).
- Desktop flows provide Robotic Process Automation (RPA) to automate legacy desktop applications by recording mouse clicks, keyboard input, and screen interactions when no API exists.
- Desktop flows can run attended (with the user signed in and watching) or unattended (on a machine without interactive sign-in), and can be triggered from a cloud flow.
- Cloud flows and desktop flows answer different problems: a cloud flow connects services that expose APIs, while a desktop flow drives the user interface of an application that does not, which is why desktop flows are the answer for legacy and terminal applications.
- Connector triggers start a flow and connector actions do the work. A trigger fires on an event - an item created in SharePoint, a response submitted to Forms, mail arriving in Outlook - and each connector publishes its own set of both.
- The Approvals action creates structured approval workflows that send requests to designated users via email and Teams; users approve or reject, and the flow branches on the outcome.
- Approval flows support sequential multi-level (hierarchical) approvals, such as team lead, then manager, then finance director.
- The Condition (if/else) action and Switch action add conditional branching so flows take different paths based on data values.
- Loop controls 'Apply to each' iterate over every item in an array or collection, and 'Do until' loops until a condition is met.
- Power Automate provides a library of pre-built templates that serve as ready-made starting points for common automation scenarios.
- Power Automate integrates deeply with Microsoft Teams to post messages, send notifications/adaptive cards, and trigger flows directly from Teams.
- Business process flows guide users through a consistent set of defined stages and steps to follow a standard business procedure (distinct from cloud and desktop flows).
- Know the named use cases: approvals routed through Outlook and Teams, notifications and adaptive cards in Teams, list and library events in SharePoint, response handling from Forms, and document automation that extracts data from invoices and forms so it can be processed without rekeying.
- AI can create and modify cloud flows from a natural-language description: state the trigger and what should happen, and the flow is drafted with the connectors and actions in place for you to adjust. It can also explain or amend an existing flow.
- AI can also create and modify desktop flows, generating the automation steps from a description of the task and helping fix a recorded flow that has broken - which matters because desktop flows are the most brittle kind to author by hand.
Domain 5: Describe Features and Capabilities of Agents in Microsoft Copilot Studio
- An agent in Copilot Studio is a configurable assistant that answers questions and takes action over your organisation's content and systems. Typical use cases are customer self-service, IT and HR helpdesks, and internal knowledge lookup - anywhere a person currently answers the same question repeatedly.
- An agent differs from a traditional chatbot in that it can reason over knowledge sources and call tools to act, rather than only following a scripted decision tree. Generative answers cover what no author anticipated; authored topics cover what must be handled a specific way.
- Topics define authored conversation paths. Each topic has trigger phrases that decide when it starts, and a set of nodes - messages, questions, conditions, variables and actions - that carry the conversation. Use a topic when the flow must be exact, such as anything with a compliance or transactional consequence.
- Questions in a topic capture user input into variables with an identified type (a date, a number, a choice), and conditions branch on those variables. Variables can be passed between topics so a value the user already gave is not asked for twice.
- Knowledge sources are what the agent grounds its generative answers on: SharePoint and OneDrive documents, public websites, Dataverse tables, and uploaded files. Scoping the knowledge is what keeps answers relevant, and the agent respects the user's existing permissions on that content.
- Generative answers are the fallback that makes an agent useful beyond its authored topics: when no topic matches, the agent answers from its knowledge sources and cites them, rather than replying that it did not understand.
- Tools are how an agent acts rather than just answers. Prebuilt and custom connectors reach other systems, and agent flows run a Power Automate process from inside the conversation - creating a ticket, updating a record, sending an approval.
- MCP servers extend an agent through the Model Context Protocol, a standard way to expose an external system's tools and data to an AI agent. Connecting an MCP server makes its whole set of tools available without building a connector for each one.
- Publish an agent to channels to reach users where they already are: Microsoft Teams, a website or custom app, Microsoft 365 Copilot, and messaging channels. Publishing is a separate step from saving, so changes are not live until the agent is republished.
- Test an agent in the test pane while authoring, where you can trace which topic was triggered and which knowledge source an answer came from. That trace is the main tool for diagnosing an agent that answers the wrong way.
- Microsoft Agent 365 is the management layer for agents across an organisation: giving agents identity, access and lifecycle management alongside the security and compliance controls applied to users, so an agent estate can be governed rather than accumulating unmanaged.
- Monitor agent usage and adoption from the analytics in Copilot Studio: how many sessions occurred, the resolution rate, escalation rate, and which topics fire most. A high escalation rate on one topic points at where the authoring or the knowledge is thin.
- Agent evaluations measure answer quality systematically rather than anecdotally: run a set of test questions against the agent and score the responses for accuracy and groundedness, so a change to topics or knowledge can be shown to have improved things rather than assumed to have.
- Agents inherit Power Platform governance: they live in an environment, are packaged in solutions for lifecycle management, are subject to data loss prevention policies on the connectors they use, and honour Dataverse security on the data they reach.
PL-900 exam tips
- Power BI is not on this exam. It is absent from the skills measured entirely, including the business-value bullets, so time spent on datasets, DAX and dashboards is time wasted. Older study material that includes a Power BI domain is out of date.
- AI Builder is not a domain any more either. The generative-AI content that replaced it is spread through the other domains: using AI to create tables and columns, to create canvas and model-driven apps, the vibe experience, and to create and modify cloud and desktop flows.
- Agents in Copilot Studio are a full 20-25% - as large as Power Apps or Power Automate. Know topics against generative answers, knowledge sources, tools including MCP servers and agent flows, publishing to channels, Microsoft Agent 365, usage analytics and agent evaluations.
- Know the canvas against model-driven distinction cold: canvas gives full layout control and connects to many data sources; model-driven is data-first, built on Dataverse, and generates its interface from the data model. Then add the two newer options - Plan designer and code apps.
- Memorise the three cloud-flow trigger types (automated on an event, instant on demand, scheduled on a recurrence) and that desktop flows are robotic process automation for applications without APIs.
- Distinguish the admin centers: the Power Platform admin center manages environments, DLP policies, capacity and analytics; the Microsoft 365 admin center manages users, groups and licences.
- Watch for true/false and "select all that apply" wording, and read carefully for distractors such as invented product names and licensing claims that sound plausible.
- Answer every question. There is no penalty for a wrong answer, and a score of 700 or greater is what you need to pass.
Study guide FAQ
How is PL-900 scored and what do I need to pass?
A score of 700 or greater is required to pass, on Microsoft's scaled 1-1000 range, and it is not a simple percentage of questions answered correctly. Microsoft states that you will have 45 minutes to complete the assessment. That is the assessment time rather than the length of the appointment, which also covers check-in and the agreements. Microsoft does not publish a fixed question count for PL-900, and it varies between forms, so prepare against the published skills rather than to a number. There is no penalty for a wrong answer, so answer every question.
Is Power BI still part of PL-900?
No. Power BI does not appear anywhere in the current skills measured, including the business-value bullets, which list Power Apps, Power Automate, Dataverse, connectors, Power Pages, generative AI features and Copilot Studio. AI Builder and the Power Pages domain were also removed, and a new domain covering agents in Microsoft Copilot Studio was added at 20-25%. Study material that still teaches a Power BI or AI Builder domain predates the 24 July 2026 update.
Do I need hands-on or coding experience to pass PL-900?
No. PL-900 is a fundamentals exam that tests conceptual understanding of what each Power Platform product does and when to use it, not the ability to build solutions or write code. Some hands-on exploration of Power Apps, Power Automate and Copilot Studio helps reinforce the concepts - the free developer environment is enough - but deep technical or coding skills are not required.
What is the difference between Power Platform and Microsoft Dataverse?
Power Platform is the suite of low-code products measured by this exam: Power Apps, Power Automate, Power Pages and Microsoft Copilot Studio. Microsoft Dataverse is the secure, cloud-based relational data platform underneath them, holding the tables, relationships, security roles and business logic those products use. Connectors are the third piece, integrating everything with data and services outside the platform.
Which licenses do I need to use connectors, and what makes a connector 'premium'?
Connectors are the integration layer between Power Platform and other data and services. Standard connectors are included with most Microsoft 365 and Dynamics 365 licences; premium connectors - such as SQL Server, Azure services and Salesforce - require a Power Apps or Power Automate premium licence; and custom connectors let an organisation wrap any REST API, including internal and legacy systems. Data loss prevention policies control which connectors may be used together, regardless of licence.
Official exam sources
The domain names and weightings on this page follow the published exam blueprint. Each source below records what it confirmed and when it was read, so the split can be checked rather than taken on trust.
- Microsoft Learn - Exam PL-900 study guideBlueprint revised July 24, 2026 · link and content verified 8 September 2026Five skills areas: 5-10 / 20-25 / 20-25 / 20-25 / 20-25, the last being Copilot Studio agents.