AI Automation for Microsoft 365:
Outlook, Teams & SharePoint

Microsoft 365 AI automation: your three options

There are three main approaches to building AI automation within a Microsoft 365 environment, each with different capabilities, costs, and technical requirements. Understanding which fits your situation is the first step.

Option 1: Microsoft Copilot (built-in AI, no setup required)

Microsoft 365 Copilot is Microsoft's native AI layer built directly into Outlook, Teams, Word, Excel, and other M365 applications. It provides: email summarisation and draft generation in Outlook, meeting transcription and action item extraction in Teams, document summarisation and analysis in Word, formula and analysis assistance in Excel, and conversational data queries in Excel and Power BI.

Copilot is the fastest path to AI assistance in M365 because it requires no workflow setup — it is activated through the Copilot button within each application. The limitation: it is manually invoked (you must be present) and does not run automated workflows that trigger on events and take actions without human involvement. Copilot enhances your manual work; it does not automate it.

Cost: Microsoft 365 Copilot adds $30/user/month to your M365 subscription. Requires M365 Business Standard or higher.

Option 2: Microsoft Power Automate (native M365 workflow automation)

Power Automate is Microsoft's native workflow automation platform, deeply integrated with M365. It supports triggers from Outlook (new email, calendar event), Teams (new message, meeting), SharePoint (new document, list item), Excel (new row), and other M365 services. It includes AI Builder, which provides pre-built AI models for document extraction, text classification, and sentiment analysis within the M365 ecosystem.

Power Automate is the right choice when: you need deep M365 integration (Teams, SharePoint, Dataverse), you work in an enterprise environment where IT must approve third-party platforms, or you need to stay entirely within the Microsoft ecosystem for security and compliance reasons.

Cost: Power Automate is included in most M365 business plans. Premium connectors and AI Builder credits require additional licensing.

Option 3: Make.com or n8n with M365 connectors (cross-platform with OpenAI)

Make.com and n8n both have native Microsoft 365 integrations — Outlook, Teams, OneDrive, SharePoint, Excel. This approach lets you combine M365 data sources with the full capability of the OpenAI API for AI processing, then send results back to M365 or any other connected system. It is the most flexible approach and typically the most capable for complex AI processing.

Cost: Make.com Core ($9/month) + OpenAI API ($5–$20/month). No additional M365 licensing required beyond your standard plan.

Automating Outlook with Make.com and OpenAI: complete walkthrough

Here is how to build the same email classification and response drafting automation described throughout this guide — but using Outlook instead of Gmail, via Make.com's native Microsoft 365 Outlook connector.

Connecting Microsoft 365 to Make.com

In Make.com, when you add a Microsoft 365 module, it uses OAuth2 authentication via your Microsoft account. In the module search, type "Microsoft 365 Email." Select "Watch Emails." Click "Add" next to Connection. Make.com opens a Microsoft authentication window — sign in with your Microsoft 365 account and grant the requested permissions (read email). The connection is stored securely and reused across all M365 modules in your scenarios.

Important for enterprise accounts: If your organisation's IT policy restricts OAuth app connections, you may need IT to approve Make.com as an authorised application in Azure Active Directory. This is a one-time approval. If Make.com requires too much IT overhead, Power Automate with a custom connector to the OpenAI API is the alternative approach within the Microsoft ecosystem.

Module sequence for Outlook email classification

Module 1 — Microsoft 365 Email "Watch Emails": Configure to watch your Inbox, limit to 5 emails per run, mark as read = No. Set schedule to every 15 minutes (free Make.com) or every 1 minute (Core plan).

Module 2 — OpenAI "Create a Completion": Connect your OpenAI API key. Model: gpt-4o-mini. System message: paste your classification prompt. User message: map the email Subject and Body Plain fields from Module 1.

Module 3 — JSON parse: Parse the OpenAI response JSON to extract category, urgency, and summary fields individually.

Module 4 — Microsoft 365 Email "Create a Draft": Create a draft reply in Outlook based on the AI's draft_reply field (if your prompt generates one). Map the original email's message ID as the reply reference.

Module 5 — Microsoft 365 Email "Move an Email" or "Add a Category": Apply a colour category to the email based on the AI's classification. Outlook categories are colour-coded labels — configure in Outlook settings before using here.

Module 6 — Google Sheets "Add a Row" (monitoring log): Log every run with timestamp, subject, classification, urgency, and draft status.

Automating Microsoft Teams with AI

Teams is the communication hub for most M365 organisations, and several high-value AI automations are possible without requiring Power Automate.

Meeting summary to Teams channel

When a Teams meeting transcript is generated, automatically post a structured AI summary to a designated Teams channel. The architecture: Microsoft Teams Webhook (when a meeting transcript is available) → Make.com → OpenAI summarisation → Microsoft Teams "Send a Message" to a specified channel. The AI summary includes: meeting overview, key decisions, action items with owners, and open questions.

Configuration note: Teams meeting transcript webhooks require a Teams App webhook subscription. This is a Power Automate flow that triggers on transcript completion and calls a Make.com webhook URL. A simpler approach: have the meeting host manually trigger the automation by saving the transcript to a watched OneDrive folder, which Make.com monitors every 15 minutes.

Alert routing from Outlook to Teams

When high-urgency emails arrive (classification = URGENT or from specific senders), automatically post an alert to a Teams channel so the relevant team member is notified immediately rather than relying on them to check Outlook. Module sequence: Outlook trigger → OpenAI classification → Router (urgency ≥ 4) → Teams "Send a Message."

Automating SharePoint and OneDrive document workflows

Document intelligence is one of the highest-value automation categories for M365 users. When new documents are uploaded to SharePoint or OneDrive, AI can automatically classify them, extract key information, generate summaries, and route them to appropriate workflows.

Document classification and tagging

Trigger: OneDrive for Business "Watch Files in a Folder" — watches a designated upload folder for new files. Processing: For Word documents and text-based PDFs, Make.com can extract the text content via Microsoft 365 modules and pass it to OpenAI for classification. For PDFs requiring visual extraction, send the file to GPT-4 Vision via the OpenAI API's file input capability. Action: Update the SharePoint document library's metadata columns with the classification, key extracted data, and a one-paragraph AI summary. This makes documents searchable and filterable by AI-extracted metadata.

Contract and proposal review automation

When a new contract or proposal PDF is uploaded to a designated SharePoint folder: extract the full text (via Microsoft 365 document processing or GPT-4 Vision for image-based PDFs), send to GPT-4o for structured extraction (parties, dates, key terms, payment conditions, renewal clauses, non-standard terms), write the extracted data back to a SharePoint list, and alert the relevant team member in Teams with a summary and link to the document. This reduces contract review time significantly for document-heavy teams.

Power Automate with OpenAI: when to use the native Microsoft platform

While Make.com often provides a better user experience and more direct access to the OpenAI API, Power Automate is the right choice in specific situations.

When Power Automate is the better choice

Enterprise security requirements: Many enterprise IT policies prohibit connecting Microsoft 365 data to third-party SaaS platforms. Power Automate is a Microsoft product that stays within the Microsoft security boundary and typically does not require third-party approval.

Deep Teams and SharePoint integration: Power Automate has first-party access to Teams and SharePoint features that third-party platforms cannot always replicate. Approvals workflows, SharePoint list item management, and Teams Adaptive Cards are all simpler in Power Automate.

AI Builder for document extraction: Microsoft's AI Builder includes pre-trained models for specific document types (invoices, receipts, business cards) that can be used in Power Automate flows without an external API key. For standard document types, AI Builder's pre-built models are faster to deploy than configuring GPT-4 Vision.

Connecting OpenAI to Power Automate

Power Automate can call the OpenAI API via its HTTP action. In your Power Automate flow, add an HTTP action with: Method = POST, URI = https://api.openai.com/v1/chat/completions, Headers = {"Content-Type": "application/json", "Authorization": "Bearer YOUR_API_KEY"}, Body = a JSON payload with your model, messages, and parameters. Parse the JSON response using Power Automate's Parse JSON action and use the extracted content in subsequent steps.

Frequently asked questions