Power Automate review

Power Automate is a Microsoft-hosted business automation service, not a general-purpose workflow library or self-hosted orchestration server. Its cloud flows automate apps, data and services in the cloud or on-premises; desktop flows use robotic process automation to operate legacy systems, desktop apps and websites through user-interface actions. The product combines those execution paths with process and task mining, Copilot-assisted authoring, and orchestration features for managing automation across an organization.

In a typical stack, Power Automate sits between business applications and the people or systems that initiate work. A flow can connect services through prebuilt or custom API connectors, while a desktop bot handles processes that still depend on a user interface. Microsoft positions it for approvals, document processing, onboarding and other business workflows. It also includes administrative capabilities such as Managed Environments, data loss prevention and monitoring. The desktop automation component makes the product broader than a cloud-only integration service, but the service remains tied to Microsoft's platform and licensing.

The product page describes an integration surface built around more than 1,000 API connectors, with the option to create custom API connectors. It also describes native Microsoft 365 experiences in Teams, Excel and SharePoint, plus cloud flows that automate apps, data and services running in the cloud or on-premises. For UI-dependent systems, desktop flows use prebuilt or custom user-interface actions rather than relying on an application connector.

The page also describes natural-language flow creation and editing with Copilot, templates, process mining and centralized administration through Managed Environments. These are useful building blocks for teams composing business workflows, but they do not replace a developer-facing API and SDK reference. The homepage does not name a programming language, SDK, CLI or protocol for building integrations. Before committing, confirm SDK support, API authentication, connector behavior and deployment tooling in Microsoft's technical documentation. Also check connector licensing and the specific actions available for each service; the count of connectors alone does not establish that a required operation is supported.

Power Automate has no free plan. The free trial lasts 30 days; the pricing details list UI-based cloud flows and standard connectors. For an individual who needs cloud flows and attended desktop automation, Power Automate Premium costs $15 per user per month, paid yearly. It includes process and task mining with 50 MB of stored data, plus Dataverse capacity of 250 MB database and 2 GB file.

Unattended desktop automation moves the bill to per-bot licensing: Power Automate Process is $150 per bot per month, paid yearly. Power Automate Hosted Process is $215 per bot per month, paid yearly, and includes a Microsoft-hosted Azure virtual machine. Both list cloud flows and Dataverse capacity of 50 MB database and 200 MB file. The Power Automate Process Mining add-on costs $5,000 per tenant per month, paid yearly, and is for Premium plans. Copilot Studio prepaid credits have custom pricing, require an Azure subscription and allow pay-as-you-go overflow. The pricing details do not publish request-based overages or rate limits, so confirm those before forecasting production costs.

Power Automate is a Microsoft product, not an open-source workflow engine. The product materials do not publish an open-source licence or offer a self-hosted Power Automate server. They describe cloud flows, desktop RPA and Microsoft-hosted infrastructure for Hosted Process; that is different from running the orchestration service on your own servers.

The product page describes flows, connectors and Microsoft Dataverse, but does not establish a general data-export or migration path. Before adopting it for critical processes, confirm how to export flow definitions and business data, and whether those exports preserve connector configuration, credentials and dependencies. Migration effort will depend on how much logic lives in flows, desktop UI actions and Microsoft-specific services. Keep business rules and source data portable where practical, document connector dependencies, and test a representative flow outside the platform before you rely on an exit plan.

Microsoft describes centralized governance, data loss prevention, monitoring and elastic scaling for hosted RPA. Those capabilities help administrators manage deployments, but teams should still validate the operational details that affect their own flows: connector and API limits, concurrency, retries, queue behavior, desktop-session availability, and how failures are surfaced. The product overview does not publish rate limits or a general reliability target, so ask Microsoft for limits and service commitments that match the workload.

Reddit reports are mixed. One thread describes onboarding and offboarding that automates most of the process, while another commenter warns that undocumented limitations can make a flow pass testing and fail in production. A separate discussion points to ownership problems when transactional workflows are built by teams that do not own the underlying process. Assign each flow an operational owner, monitor failure paths, and test changes against production-like data and permissions. Treat desktop UI automation as especially dependent on the stability of the target application's interface.