What is the Power Platform?

The Microsoft Power Platform is a low-code application platform that pairs a suite of maker products with a shared data foundation. The maker products are Power Apps (business applications), Power Automate (workflow automation), Power BI (analytics), Power Pages (external-facing web experiences), and Copilot Studio (conversational AI agents). The shared foundation is Microsoft Dataverse — a relational, secure, enterprise-managed data service.

Dynamics 365's CRM-lineage apps (Sales, Customer Service, Field Service, Marketing, Customer Insights, Project Operations) are themselves built on the Power Platform. They ship as pre-configured Power Apps on top of Dataverse. This is not marketing framing — it is architecturally true — and it is why extending Dynamics 365 uses exactly the same tools a maker would use for a fresh Power Apps build.

The five maker products

Power Apps builds business applications in two flavours. Model-driven apps generate a functional data-centric UI from Dataverse metadata — the pattern Dynamics 365 CRM apps use, right for record-heavy processes with security roles and complex forms. Canvas apps are drag-and-drop pixel-perfect designs backed by a spreadsheet-like formula language (Power Fx), right for task-specific workflows that need a bespoke UI.

Power Automate runs workflows. Cloud flows trigger from events (a record change in Dataverse, an email arriving, a form submitted, a schedule) and run steps against 500+ connectors. Desktop flows are RPA — a bot that drives a Windows UI when there is no API — increasingly overtaken by the API-first world but still important for legacy systems. Business process flows are the guided-process flavour used inside model-driven apps to shepherd a user through Qualify → Develop → Propose stages.

Power BI is the enterprise BI product — semantic models over Fabric, Dataverse, SQL, files, everything. It sits inside the Power Platform brand but is really its own product with its own licence economy and its own admin surface; the overlap is that Dataverse is a first-class data source and that Power BI reports embed into Power Apps and dashboards.

Power Pages builds external-facing websites — customer portals, partner portals, self-service forms — with Dataverse as the back end and its own authentication (Entra External ID, Azure B2C, local accounts). It is the descendant of the older Power Apps Portals product, rewritten in a modern design experience.

Copilot Studio builds conversational AI agents — chatbots and Teams-native agents that reason over instructions, tools, and knowledge. It is the successor to Power Virtual Agents; the redesign around Copilot happened in 2024 and continues.

Dataverse

Dataverse sits under the maker products. It is a relational data service with an enterprise-grade security model (role-based, row-level, hierarchical, field-level), a rich metadata layer (tables, columns, relationships, choice sets, business rules), and a scalable transactional storage engine.

Every Power Platform environment ships with a Dataverse instance. A Dynamics 365 tenant's Dataverse is the same Dataverse that hosts custom Power Apps — with the pre-built Dynamics 365 tables layered on top. That means a custom Power App can add a table that relates to the Account table and it will show up on Account forms in Sales — no integration involved.

The extended Dataverse family includes elastic tables (NoSQL-style storage for high-volume append workloads), virtual tables (federated data staying in its source system), file/image columns (blob storage attached to records), and Dataverse for Teams (a lightweight Dataverse that ships free inside Teams). All share the same programming model.

Environments and governance

A Power Platform tenant is organised into environments — logically separated containers, each with its own Dataverse, security roles, users, and app inventory. The typical setup is a development environment, one or more test environments, and production, mirroring the classic ALM structure. A default environment exists for personal productivity apps that shouldn't touch enterprise data.

Managed environments is a paid tier that adds sharing controls, weekly usage reports, solution checker enforcement, and pipelines. It's designed for organisations that want central IT visibility over a wide maker community, not for a small dev-only footprint.

Data Loss Prevention (DLP) policies categorise every connector as Business, Non-Business, or Blocked, and prevent a flow or app from combining connectors across categories. This is the primary governance lever — set it right and low-code makers can't accidentally exfiltrate Dataverse data to a personal Dropbox.

ALM

Power Platform applications ship as solutions — packaged units of tables, apps, flows, security roles, and dependencies that move between environments. Solutions come in two flavours: unmanaged (editable in the destination environment, used in dev) and managed (locked in the destination, used in test and production).

Power Platform Pipelines is the newest and simplest deployment path — a checklist-driven promotion from dev to prod inside the Power Platform admin center. For heavier ALM there is Power Platform Build Tools for Azure DevOps and GitHub Actions for Power Platform, both wrapping the same pac CLI. Larger tenants that want a full ALM discipline use the CLI + Actions or DevOps combination.

The AI layer

Every product in the platform has a Copilot inside it. Copilot in Power Apps turns natural-language descriptions into apps, tables, and formulas. Copilot in Power Automate does the same for flows. Copilot in Power BI authors DAX and describes visuals. Copilot Studio is itself the AI-agent building product.

Underneath, AI Builder provides pre-built AI models (form processing, sentiment, text translation, custom prediction) that flows and apps can call as steps — a lower-code alternative to calling Azure Cognitive Services directly.

Licensing shape

Licensing is intricate and changes frequently. In broad strokes: Power Apps is licensed per app or per user, Power Automate per user or per flow, Power BI per user (Pro, Premium Per User) or per capacity, Power Pages per site with anonymous/authenticated user metering, Copilot Studio per message pack. Dataverse consumption (storage and API calls) is metered against the tenant.

Every Dynamics 365 licence includes rights to use the CRM-lineage app's own Dataverse tables inside custom apps and flows. Extending Dynamics 365 with a custom app that uses only Dynamics tables typically does not need an additional Power Platform licence; a custom app that adds new Dataverse tables usually does. Always check the licensing guide before assuming.

Where the platform starts and ends

Power Platform is a low-code platform. It optimises for time-to-first-app, integration breadth, and delegation to Dataverse's transactional engine. It is not a pro-code IDE, not a raw ML platform, and not a general-purpose Kubernetes runtime — for those, Azure sits alongside it (Azure Functions, Azure ML, AKS).

The right mental model: Power Platform is the delivery layer for business applications and automations that need to be built quickly, governed centrally, and integrated with Microsoft 365 and Dynamics 365. Azure is the platform for everything more ambitious than that.

Choosing the first product

New teams routinely try to adopt all five at once and stall. In practice the entry point is the shape of the first problem, not a platform-wide rollout:

Shape of the first problem	Start with	Why
A shared form or lightweight app replacing a spreadsheet	Power Apps (canvas) with Dataverse	Fast time-to-first-app, security at the row level, no server
A record-heavy business process with roles and views	Power Apps (model-driven) with Dataverse	The pattern Dynamics 365 CRM apps use; scales with the process
An approval, notification, or scheduled data movement	Power Automate cloud flow	Trigger + connectors solves most SaaS integration without code
A UI-driven task against a legacy Windows app with no API	Power Automate desktop (RPA)	Bridging patterns while an API-first replacement is being built
A report or dashboard for the business	Power BI	Semantic model, distribution, phone-friendly visuals
A public sign-up, application, or customer self-service	Power Pages	Anonymous + authenticated pages backed by Dataverse
A conversational agent in Teams or on a site	Copilot Studio	Instructions + tools + knowledge without a bot framework

The rule of thumb: pick one product for the first delivery, land it in production with real users, then add the second product once the first is stable. Trying to master all five in parallel is what makes governance a mess later.

First-time buyer checklist

Before signing a Power Platform commitment or handing an environment to a first business team:

• Environment strategy. How many environments (dev, test, prod, default), who can create them, and what does the default environment hold? See Power Platform environments.
• DLP policies. At minimum: Business + Non-Business classification of connectors, a Blocked set for personal storage and consumer social, one policy per environment tier. See DLP policies in Power Platform.
• Identity and licensing. Which users have Power Apps rights today via Microsoft 365 (limited), which need per-user or per-app; who administers the Power Platform admin center and Managed Environments.
• Data source. Dataverse is the default and cheapest to govern; SharePoint lists and Excel look free but are the biggest source of ungoverned sprawl. Decide the rule.
• ALM path. Pipelines for the simple case; Power Platform Build Tools or GitHub Actions once solutions cross environments daily. See ALM with GitHub Actions for Power Platform.
• Governance visibility. Center of Excellence Starter Kit or Managed Environments reports — one of them needs to be watching maker activity from day one, not after month twelve.
• Copilot posture. Whether makers can use Copilot in Power Apps and Power Automate, which connectors it can reason over, and whether Copilot Studio agents are allowed to be published outside Teams.
• Sunset plan. How an app or flow that lost its owner gets identified and retired — the single most under-discussed piece of adoption.

Where to go next

Next reads: Power Platform environments for the operating model, canvas apps vs model-driven apps for the first build decision, flow design patterns for automation, DLP policies for governance, and Copilot Studio for Dynamics 365 for agents. Licence costs are on the Power Apps and Power Automate pricing pages.