9.1 Cloud Flow Types: Automated, Instant, Scheduled & Desktop Overview
Key Takeaways
- Cloud flows are divided into three architectural trigger types: Automated (event-driven/webhooks/polling), Instant (manual button triggers from Power Apps, SharePoint, mobile, or Dataverse), and Scheduled (recurrence patterns with time-zone precision).
- Instant flows support granular user input parameters (Text, Date, Number, Boolean, File) and expose Run-only user permissions to determine whether actions execute under maker credentials or the invoking user's connection.
- Scheduled flows evaluate recurrence using ISO 8601 UTC timestamps, offering day-of-week, hour-of-day, and minute-level precision with native Daylight Saving Time (DST) handling.
- Robotic Process Automation (RPA) via Power Automate for Desktop bridges legacy desktop and web applications lacking native APIs, orchestrated seamlessly from cloud flows.
- Attended RPA executes on an active user desktop sharing human input controls, whereas Unattended RPA executes autonomously on dedicated virtual machines in locked, automated user sessions.
Cloud Flow Types: Automated, Instant, Scheduled & Desktop Overview
Microsoft Power Automate is the enterprise workflow automation and process orchestration service within the Microsoft Power Platform. Built atop Azure Logic Apps infrastructure, Power Automate provides an intelligent, low-code engine capable of connecting hundreds of cloud services, on-premises systems, and legacy desktop applications. For the PL-200: Microsoft Power Platform Functional Consultant certification exam, you must master the fundamental cloud flow trigger types, understand their underlying execution models, configure manual and automated runtime parameters, and architect robotic process automation (RPA) orchestrations bridging modern cloud APIs and legacy client software.
1. Power Automate Architecture & Cloud Flow Taxonomy
Every cloud flow begins with a single Trigger—the catalyst that instantiates the workflow pipeline. Once triggered, the flow executes an ordered sequence of Actions (such as querying records, executing business logic, branching conditionally, or sending notifications).
+-----------------------------------------------------------------------------+
| POWER AUTOMATE CLOUD FLOW TAXONOMY |
| |
| +---------------------------------------------------------------------+ |
| | POWER AUTOMATE ENGINE | |
| | (Azure Logic Apps Runtime) | |
| +---------------------------------------------------------------------+ |
| | | | |
| v v v |
| +-----------------------+ +-----------------------+ +-----------+ |
| | AUTOMATED FLOWS | | INSTANT FLOWS | | SCHEDULED | |
| | - Event-Driven | | - Manual Button Push | | - Time / | |
| | - Push (Webhooks) | | - Mobile / Portal | | Interval| |
| | - Polling (Interval) | | - Power Apps (V2) | | - Recur | |
| | - Dataverse/SharePoint| | - 'For selected item' | | - Specific| |
| | create/update/delete| | - Run-only credentials| | Days/Hrs| |
| +-----------------------+ +-----------------------+ +-----------+ |
| | | | |
| +----------------------------+----------------------------+ |
| | |
| v |
| +------------------------------------------+ |
| | DESKTOP FLOW ORCHESTRATION (RPA) | |
| | - Attended: User logged in, interactive | |
| | - Unattended: Autonomous VM, locked session| |
| | - Direct Machine / Gateway Connectivity | |
| +------------------------------------------+ |
+-----------------------------------------------------------------------------+
Power Automate classifies flows into distinct operational patterns:
- Automated Cloud Flows: Triggered automatically by an external event in a connected system (e.g., a new email arrives, a Dataverse row is modified, or a SharePoint item is created).
- Instant (Button) Cloud Flows: Triggered on-demand by a human user or external client application through a manual button press (e.g., from the Power Automate mobile app, a Canvas app via PowerApps V2 trigger, or a context menu in SharePoint/Dataverse).
- Scheduled Cloud Flows: Triggered automatically at predefined calendar recurrence intervals (e.g., daily at 06:00 AM, every Monday and Thursday, or every 15 minutes).
- Desktop Flows (RPA): Automated UI scripts running on Windows workstations or virtual machines, recording keyboard/mouse actions to drive legacy applications lacking REST APIs.
2. Automated Cloud Flows & Event-Driven Triggers
Automated flows execute in direct response to system events. Behind the scenes, automated triggers operate using one of two fundamental communication protocols:
+-----------------------------------------------------------------------------+
| PUSH (WEBHOOK) VS POLLING TRIGGERS |
| |
| [PUSH / WEBHOOK TRIGGER] [POLLING TRIGGER] |
| - Event occurs in source system - Power Automate polls endpoint |
| - Source calls flow HTTP callback - Runs at configured interval |
| - Near-instantaneous execution (<1 sec) - Latency: 1 to 15 minutes |
| - Example: Dataverse 'When a row is - Example: SQL 'When an item is |
| added, modified, or deleted' created' (legacy polling) |
+-----------------------------------------------------------------------------+
Push (Webhook) Triggers
In a push architecture, the target cloud service registers an asynchronous webhook subscription with Power Automate. When an event occurs (e.g., a customer submits a Microsoft Form or a record is created in Microsoft Dataverse), the service immediately dispatches an HTTP POST request to the flow's unique callback endpoint. Execution is near-instantaneous (sub-second latency).
Polling Triggers
In a polling architecture, the Power Automate engine makes recurring HTTP GET requests to the target service API at fixed intervals (typically every 1 to 15 minutes depending on environment licensing) to check for new or modified records. The trigger maintains a state cursor (such as a high-water mark timestamp or delta token) to detect items created since the previous poll.
Concurrency Control & Split On
When a trigger returns an array of records (e.g., a polling trigger that detects 25 new emails simultaneously), Power Automate uses the Split On mechanism:
- Split On Enabled (Default): The engine splits the incoming array and instantiates 25 separate, independent flow runs in parallel.
- Split On Disabled: The flow executes a single instance, passing the entire array of 25 records into the trigger body, requiring an
Apply to eachloop inside the flow to iterate through them. - Concurrency Control: In the trigger's Settings pane, makers can enable Concurrency Control and set the Degree of Parallelism (1 to 50). Setting parallelism to
1forces sequential execution, guaranteeing that records are processed one at a time in strict arrival order.
3. Instant (Button) Cloud Flows & Manual Triggers
Instant flows are designed for scenarios where human judgment or manual intervention dictates when a process should start.
+-----------------------------------------------------------------------------+
| INSTANT FLOW TRIGGER ECOSYSTEM |
| |
| [MANUALLY TRIGGER A FLOW] [POWERAPPS (V2) TRIGGER] |
| - Power Automate Mobile App - Canvas Apps & Custom Pages |
| - Web Portal Button - Power Fx: FlowName.Run(...) |
| - User input prompt parameters - Strongly typed input schemas |
| |
| [FOR A SELECTED ITEM (SHAREPOINT)] [FOR A SELECTED ROW (DATAVERSE)] |
| - SharePoint Document Library/List - Model-Driven Grid/Form Command |
| - Contextual: ID, FilePath, User - Contextual: EntityName, GUID |
+-----------------------------------------------------------------------------+
User Input Parameters in Manual Triggers
When configuring a Manually trigger a flow step, makers can define dynamic input prompts that require users to provide data before execution begins:
- Text: Free-form single-line or multi-line string.
- Drop-down list (Choice): Predefined array of valid text options.
- Date: Calendar date selector (ISO format
YYYY-MM-DD). - Number: Integer or floating-point numerical input.
- Yes/No (Boolean): Toggle switch.
- File: Base64-encoded binary payload (e.g., photo taken on a smartphone camera or document attachment).
Instant Flow Host Triggers
| Trigger Name | Host Environment | Contextual Payload Generated |
|---|---|---|
| Manually trigger a flow | Power Automate Mobile App, Flow Web Portal | Maker-defined input parameters, User email, Timestamp, Latitude/Longitude (mobile) |
| PowerApps (V2) | Canvas Apps, Custom Pages | Strongly typed parameters passed via Power Fx FlowName.Run(Param1, Param2) |
| For a selected item | SharePoint Online Lists & Document Libraries | ID, ListUrl, ItemUrl, UserEmail, user input parameters |
| For a selected row | Dataverse Model-Driven App Command Bar | entityName, entityId (Record GUID), userId, user input parameters |
Run-Only Users & Connection Security
For instant flows shared across organizational teams (such as button flows or SharePoint contextual flows), the maker must configure Run-only users permissions (Flow Details > Edit Run-only users):
+-----------------------------------------------------------------------------+
| RUN-ONLY USER PERMISSION ARCHITECTURES |
| |
| [USE THIS CONNECTION (FLOW OWNER / SERVICE ACCOUNT)] |
| - Flow executes using creator/service credentials |
| - End users do NOT need direct security permissions to backend data store |
| - Safe for writing to restricted audit logs or elevated approval steps |
| |
| [PROVIDED BY RUN-ONLY USER (INTERACTIVE CREDENTIALS)] |
| - Flow executes using the calling user's own identity and OAuth token |
| - Respects calling user's row-level security and Dataverse privileges |
| - Backend actions fail if the run-only user lacks target permissions |
+-----------------------------------------------------------------------------+
[!IMPORTANT] PL-200 Exam Rule for Run-Only Users: If an instant flow must write an audit record to a central SharePoint list or Dataverse table that general frontline employees are forbidden from accessing directly, configure the connection under Run-only users to "Use this connection (Owner)". If the flow must send emails appearing from the employee's personal mailbox or create records tagged with their identity, configure the connection as "Provided by run-only user".
4. Scheduled Cloud Flows & Time-Based Recurrence
Scheduled cloud flows execute automatically according to a predefined time schedule. They are configured using the built-in Recurrence trigger.
+-----------------------------------------------------------------------------+
| RECURRENCE TRIGGER PARAMETERS |
| |
| - Interval: 1, 2, 15, 30 ... (Positive integer) |
| - Frequency: Month, Week, Day, Hour, Minute, Second |
| - Time Zone: (UTC-05:00) Eastern Time (US & Canada) |
| - Start Time: 2026-08-17T06:00:00Z (ISO 8601 UTC format) |
| - On these days: Monday, Wednesday, Friday (When Frequency = Week) |
| - At these hours: 6, 18 (When Frequency = Day or Week) |
| - At these minutes: 0, 30 (When Frequency = Day, Week, or Hour) |
+-----------------------------------------------------------------------------+
Advanced Recurrence Configuration
When Frequency is set to Week or Day, advanced scheduling properties appear:
On these days: Select specific calendar days (e.g.,Monday, Wednesday, Friday).At these hours: Specify execution hours in 24-hour military format (e.g.,8, 12, 17for 8:00 AM, 12:00 PM, and 5:00 PM).At these minutes: Specify exact minute marks within the hour (e.g.,0, 30for top and bottom of the hour).- Time Zone & Daylight Saving Time: Selecting an explicit named Time Zone (e.g.,
(UTC-05:00) Eastern Time (US & Canada)) ensures the Power Automate scheduler automatically compensates for Daylight Saving Time (DST) shifts. If no time zone is selected, the trigger defaults to UTC without local clock adjustments.
[!TIP] Avoiding Schedule Drift: Always specify both a
Start time(e.g.,2026-01-01T08:00:00Z) and an explicitTime zone. If a flow frequency is set toDaywith an interval of1without a start time, the flow runs exactly 24 hours after it was saved, drifting over time due to slight execution latencies.
5. Power Automate for Desktop (RPA) Overview & Cloud Orchestration
While cloud flows connect modern systems via cloud APIs, many enterprise operations depend on legacy Windows client applications, AS/400 terminal emulators, or web portals lacking REST APIs. Power Automate for Desktop (PAD) provides Robotic Process Automation (RPA) capabilities to automate these UI interactions.
+-----------------------------------------------------------------------------+
| CLOUD-TO-DESKTOP FLOW ORCHESTRATION PIPELINE |
| |
| [CLOUD FLOW TRIGGER] (Automated / Instant / Scheduled) |
| | |
| v |
| [ACTION: Run a flow built with Power Automate for Desktop] |
| | |
| +---> Passes Input Variables (JSON, String, Numbers) |
| | |
| v |
| [DESKTOP AGENT CONNECTIVITY] |
| - Direct Machine Connectivity OR On-Premises Data Gateway |
| - Machine Group (Dynamic load balancing across VM pool) |
| | |
| +------------------------------------+ |
| | | |
| v v |
| [ATTENDED RPA] [UNATTENDED RPA] |
| - User is actively logged in - Autonomous dedicated VM / session |
| - Shares screen & input controls - Signs into Windows automatically |
| - Real-time user supervision - Screen locked during execution |
| - Desktop agent runs interactively - Signs out when execution completes |
| | | |
| +------------------------------------+ |
| | |
| v |
| [RETURNS OUTPUT VARIABLES TO CLOUD FLOW] |
| - Execution status, extracted tabular data, error codes |
| | |
| v |
| [SUBSEQUENT CLOUD FLOW ACTIONS] (Update Dataverse, Send Email, etc.) |
+-----------------------------------------------------------------------------+
Attended vs. Unattended RPA
| Architectural Dimension | Attended RPA | Unattended RPA |
|---|---|---|
| User Session State | User is actively signed into Windows workstation | No user logged in; system creates/destroys automated Windows session |
| Initiation Method | Triggered by user action on their desktop or an instant cloud flow | Triggered automatically by scheduled or event-driven cloud flows |
| Workstation Interaction | Uses the user's active screen, keyboard, and mouse | Runs headlessly on dedicated physical machine or Azure Virtual Machine |
| Screen State | Screen must remain unlocked and visible to user | Screen is locked / headless; runs securely without user monitoring |
| Target Use Cases | Frontline customer service, call center assisted tasks | High-volume nightly batch data entry, invoice extraction, ledger posting |
| Licensing Model | Power Automate Premium user license | Power Automate Process license / Hosted RPA add-on per bot/machine |
| Machine Orchestration | Single target workstation | Machine Groups with dynamic queueing and automated load balancing |
Cloud-to-Desktop Flow Orchestration Mechanics
Cloud flows invoke desktop flows using the action Run a flow built with Power Automate for Desktop:
- Connection Mode: Connects to desktop machines via Direct Machine Connectivity (recommended in modern environments) or an On-Premises Data Gateway.
- Machine Groups: For high-throughput unattended workloads, machines are grouped into a Machine Group. The cloud action targets the group, and Power Automate dynamically routes the job to the first available idle machine.
- Variable Passing: Input variables defined in the Desktop Flow (e.g.,
CustomerTaxID,InvoiceTotal) are populated directly by dynamic content from the cloud flow. Upon completion, desktop output variables (e.g.,ConfirmationCode,ProcessingStatus) are returned to the cloud flow pipeline for downstream processing.
A functional consultant is designing an instant cloud flow triggered from a SharePoint document library. Frontline team members will click the flow button to submit completed invoices for executive sign-off. The flow must post an entry into a central, restricted Dataverse audit table that frontline users are strictly forbidden from accessing directly, and then send an approval email from a corporate service mailbox. How should the consultant configure the flow's Run-only users settings?
An enterprise organization requires a scheduled cloud flow to generate and distribute an executive compliance digest every Tuesday and Thursday at exactly 07:00 AM and 03:00 PM Central Standard Time. The consultant must ensure the schedule automatically respects daylight saving time transitions without time drifting over months of operation. How should the Recurrence trigger be configured?
An enterprise bank processes nightly batch loan reconciliations by reading data from an AS/400 green-screen mainframe terminal emulator. The batch job starts at 01:00 AM every night on a pool of Azure Virtual Machines where no human operator is logged into the Windows console. The process must dynamically distribute execution across the VM pool and run in a locked, secure session. Which automation architecture should the consultant implement?
A cloud flow is triggered by the automated trigger 'When a new email arrives (V3)'. During peak morning hours, the corporate mailbox receives batches of up to 40 customer inquiry emails simultaneously. By default, the flow engine batches these incoming emails and processes them sequentially in a single instance, leading to significant delays. The consultant needs each received email to immediately trigger its own separate, concurrent flow run. Which trigger setting must be enabled?