ChatGPT Prompts for Power Automate Flows
20 copy-paste ChatGPT prompts for Power Automate: flow design, approval workflows, integrations across Microsoft 365, error handling, and the RPA work that shifts org from manual to automated.
In short: This page contains 16 copy-paste ready prompts, organized into 4 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Hand-curated and tested by the AI Academy team.
Flow Design
4 promptsFlow Design from Process
1/16✨ What it does
ChatGPT turns [Paste manual process] into a Power Automate flow with trigger, action sequence, branches, notifications, error handling, and named connectors. Build the flow from that map and run it once on a safe test item you control.
[Paste manual process]. Design Power Automate flow. Output: trigger (event-based vs scheduled vs manual), action sequence, conditions + branches, output/notifications, error handling. Walk through logically; specific connectors named.
Pro tip: Manual processes documented = automation candidates. AI helps design flow logic; you build in Power Automate. Translate process steps to flow steps systematically.
Approval Flow Design
2/16✨ What it does
ChatGPT designs an approval flow for [process], covering trigger, sequential or parallel approvers, form fields, rejection, escalation, notifications, audit, and the completion action. Name the approvers and send one test request through so you see the path.
Approval flow for [process]. Output: trigger, approvers (sequential vs parallel vs custom), approval form fields, rejection path, escalation, notifications, audit log, completion action. Approvals = most common automation.
Pro tip: Approval flows replace email-chain approvals. Audit trail + notifications + escalation built-in. Email approval = lost; flow approval = tracked.
Connector Selection
3/16✨ What it does
ChatGPT picks connectors for [scenario integrating systems X and Y], covering Microsoft versus certified versus community, premium license cost, auth, limits, and a fallback. Check the license on the connector before you build the flow.
Connectors for [scenario integrating systems X and Y]. Output: best connector option (Microsoft / certified / community), license requirements (premium connectors cost), authentication method, common limitations, alternative if connector limited.
Pro tip: Premium connectors require per-user license. Standard connectors free. Many flows can use HTTP REST as fallback for missing connectors. Plan license cost.
Trigger Strategy
4/16✨ What it does
ChatGPT chooses a trigger for [scenario] among scheduled, event, manual, HTTP, and Teams instant, with why, alternatives, and common issues. Create the flow with that trigger and fire it once to confirm it starts when you expect.
Best trigger for [scenario]. Options: scheduled (recurrence), event (when item created/modified), manual (button), HTTP, instant (Teams). Output: recommended trigger, why, alternatives, common issues. Trigger choice = flow reliability.
Pro tip: Wrong trigger = flow doesn't run when needed or runs too often. Event triggers more responsive than scheduled; scheduled more predictable. Match to use case.
Prompts get you started. Tutorials level you up.
A growing library of 300+ hands-on AI tutorials. New tutorials added every week.
Common Automations
4 promptsNew Email → SharePoint
5/16✨ What it does
ChatGPT designs a flow that saves matching email attachments and bodies to SharePoint using [criteria], with sender or subject conditions, file path, metadata, and notifications. Test with one email you control, then turn the flow on.
Flow: when email arrives matching [criteria], save attachment + email body to SharePoint with metadata. Output: trigger conditions (sender / subject / has-attachment), file path logic, metadata mapping, notifications. Email-to-SP automation = inbox decluttered.
Pro tip: Common pattern: invoices/POs/contracts via email → archive in SharePoint. Replaces manual save+rename process. Compounds time savings across team.
Form → Notification Flow
6/16✨ What it does
ChatGPT designs a flow where a Microsoft Form submission sends a Teams ping and logs a row in Excel, including parsing, a condition, the append, and error handling. Submit a dummy form and confirm both the ping and the row in your log.
Flow: Microsoft Form submission triggers Teams notification + Excel logging. Output: form-data parsing, conditional notification (only if X), Excel append, error handling. Form responses don't notify by default.
Pro tip: MS Forms responses sit in form analytics. Notification flow = team responds in real-time. Excel logging = analysis layer. Zero default; high value to add.
Recurring Task Reminder
7/16✨ What it does
ChatGPT builds a weekly reminder flow for [recurring task], with schedule, recipients, message text, and an escalation if completion is not tracked. Turn it on for one task and mark completion in the form or check your team uses.
Flow: weekly reminder for [recurring task]. Output: scheduled trigger, recipient list, message customization, escalation if not completed (track completion via Forms or check). Beat the "I forgot" excuse.
Pro tip: Email reminders ignored. Teams pings + form-based "did you do it?" + escalation = compliance. Reminder design matters.
Document Approval + Notification
8/16✨ What it does
ChatGPT designs a SharePoint document approval flow that triggers on new or changed files, routes to an approver group, stores the decision, and notifies the author. Upload a test file and watch the approval land, then check your metadata.
Flow: SharePoint doc requires approval. Output: trigger on new/changed, route to approver group, parallel sign-offs if needed, store decision metadata, notify originator. SharePoint built-in approvals weak; PA replaces.
Pro tip: SharePoint native approval limited. Power Automate approval = parallel approvers + escalation + notifications + audit. Standard for serious approval needs.
Error Handling + Maintenance
4 promptsError Handling Pattern
9/16✨ What it does
ChatGPT adds error handling to flow [describe], with configure run after, failure notice, retry, a fallback, and logging so failures are not silent. Force a bad step in a copy of the flow and confirm the notice reaches you.
Add error handling to flow [describe]. Output: try-catch pattern (configure run after settings), failure notification, retry logic, fallback action, logging to track recurring issues. Default flows fail silently; designed flows surface errors.
Pro tip: Default flow failure = silent. Run-after configuration on actions = catch failures + alert. Production flows need error handling; prototype flows skip.
Flow Performance Optimization
10/16✨ What it does
ChatGPT optimizes [Paste flow description] by cutting extra actions, batching, filtering early, using parallel branches, and lighter actions so the run does not time out. Apply the cuts on a copy and compare run duration before you publish.
[Paste flow description]. Optimize: minimize action count, batch operations, filter early (don't pull all then filter), parallel branches where possible, choose lighter actions. Slow flows = time out + fail.
Pro tip: Each action adds latency. 50-action flow slow + fragile. 10-action flow fast + reliable. Optimization = production stability.
Flow Documentation
11/16✨ What it does
ChatGPT documents flow [name] for handoff, with purpose, trigger, step-by-step actions, inputs and outputs, error cases, owner, last update, and dependencies. Store that page next to the flow and name a backup owner besides you.
Document flow [name] for handoff/maintenance. Output: purpose, trigger, action sequence (step-by-step), inputs/outputs, error scenarios + handling, owner, last update, dependencies. Undocumented flows = unmaintainable.
Pro tip: Flow author leaves = flow becomes mystery. Quarterly docs update = sustainable flow library. Most orgs let flows decay; documentation is the prevention.
Flow Audit + Cleanup
12/16✨ What it does
ChatGPT plans an audit of [tenant's] Power Automate flows, listing what runs, what fails, owners, value versus cost, retirement candidates, and license use. Turn off one unused flow after you confirm nobody still needs it.
Audit [tenant's] Power Automate flows. Output: flows running, flows failing, owner mapping, business value vs cost, candidates for retirement, license utilization. Flow sprawl = real cost.
Pro tip: Flow sprawl in 2 years = 100s of flows, 50% inactive, 25% failing. Annual audit + retirement = sustainable. Without audit = invisible cost accumulation.
Like these prompts? There are full tutorials behind them.
Learn the workflows, not just the prompts. 300+ easy-to-follow tutorials inside AI Academy — and growing every week.
Advanced Patterns
4 promptsHTTP Request + JSON Parse
13/16✨ What it does
ChatGPT designs a flow that calls API [endpoint], with HTTP method, URL, headers, auth, JSON parse, API error handling, and rate limits. Test against a sandbox endpoint before you point it at production.
Flow calls API: [endpoint]. Output: HTTP action setup (method, URL, headers, auth), JSON parse for response, error handling for API failures, rate limit handling. APIs without connectors = HTTP fallback.
Pro tip: Most SaaS APIs have no Power Automate connector. HTTP action = generic fallback. Authentication + parsing learned once = unlocks everything API-callable.
Loop Through Array
14/16✨ What it does
ChatGPT sets up Apply to each for [describe], covering parallel versus sequential, the per-item action, aggregation, and performance on large arrays. Run it on a short list first, then raise concurrency only if your volume needs it.
Flow needs to process array of items: [describe]. Output: Apply to Each setup, parallel vs sequential decision, action per item, aggregation if needed, performance considerations for large arrays. Arrays = common automation pattern.
Pro tip: Apply to Each = workhorse. Default sequential; parallel option in settings (5x speedup). Large arrays = batch processing pattern (slice into chunks).
Power Automate Desktop (RPA)
15/16✨ What it does
ChatGPT compares Power Automate Desktop versus cloud flows for [scenario], covering when a legacy GUI needs desktop RPA, when cloud is enough, a hybrid pattern, and licenses. Pick cloud unless the app has no API, then note the license you will need.
PAD vs cloud flows for [scenario]. Output: when desktop RPA needed (legacy app, no API, GUI automation), when cloud sufficient, hybrid pattern (cloud triggers desktop), license requirements. RPA last resort.
Pro tip: Desktop RPA fragile (UI changes break it). Cloud flows + API > RPA when possible. RPA when no other option (legacy systems with no API). Don't default to RPA.
Custom Connector
16/16✨ What it does
ChatGPT plans a custom connector for [API], covering OpenAPI research, auth, actions, parameter mapping, testing, and sharing with the org. Test each action on the connector before any production flow you own calls it.
Build custom connector for [API]. Output: API research (OpenAPI spec ideal), authentication setup, action definitions, parameter mapping, testing, sharing with org. Custom connectors = reusable across flows.
Pro tip: Common API used in many flows = build custom connector once. Saves repeated HTTP setup. Sharing across org = reuse compounds.
Free tool
Prompt Optimizer
Turn a rough idea into a structured, professional AI prompt.
Frequently Asked Questions
Prompts are the starting line. Tutorials are the finish.
A growing library of 300+ hands-on tutorials on ChatGPT, Claude, Midjourney, and 50+ AI tools. New tutorials added every week.
7-day free trial. Cancel anytime.
Related guides