Bring the full power of Microsoft Outlook into your Gumloop workflows. Search, send, and organize emails with plain-language prompts so you never have to switch tabs.
Work with every folder in your Outlook mailbox - Inbox, Drafts, Archive, and more - directly from Gumloop.

​
What is Outlook MCP?

Outlook MCP creates a customized node that speaks Microsoft Outlook’s API language, giving you AI-powered access to your mailbox through simple, natural-language prompts.

​
What Can It Do for You?

  • Instantly retrieve emails that match specific senders, subjects, or dates
  • Send polished messages or schedule follow-ups without opening Outlook
  • Keep your inbox tidy by archiving, deleting, or flagging messages on command
  • Draft emails ahead of time, then update or forward them whenever you’re ready

​
Available Tools

ToolWhat It DoesExample Use
Read EmailsFetches emails from Outlook based on filters like sender, subject, folder, or date range.”Read emails from john.smith@acme.com received this week and return subject, sender, receivedDateTime”
Send EmailSends a new email to one or more recipients.”Send an email to jenny@contoso.com with subject ‘Project Kickoff’ and body ‘[message_content]’ and return messageId”
Update EmailChanges properties of an email such as read/unread status, flagged status, or moves it to a different folder.”Mark all emails from noreply@github.com as read and return id, isRead”
Create DraftCreates a draft message in any folder.”Create a draft to alex@alpha.io with subject ‘Quarterly Report’ and body ‘[draft_content]’ and return draftId, subject”
Forward EmailForwards an existing email to specified recipients with an optional comment.”Forward the latest email with subject ‘Budget Approval’ to sara@beta.com adding comment ‘Please review by EOD’ and return forwardId”
Archive EmailMoves an email to the Archive folder.”Archive all emails from newsletter@medium.com older than 30 days and return id, folder”
Trash EmailMoves an email to Deleted Items.”Move emails with subject containing ‘Unsubscribe Confirmation’ to trash and return id, folder”

​
How to Use

1

Create Your Outlook MCP Node

Go to your node library, search for Outlook, and click “Create a node with AI”
2

Add Your Prompt

Drag the Outlook node onto the canvas and type a clear, single-action prompt such as “Read my unread emails from yesterday.”
3

Test Your Node

Run the node to check the output. Adjust your prompt or filters until you see the exact data you need.
4

Save and Reuse

Once everything looks good, save the node to reuse it in any future workflow.

​
Example Prompts

Here are some prompts that work well with Outlook MCP: Retrieve Messages: 
Read emails from finance@acme.com received since 2025-01-01 and return subject, sender, receivedDateTime
Send a New Email: 
Send an email to marketing@contoso.com using the provided subject and email body input and return messageId
Create a Draft: 
Create a draft email to tom@startups.io with subject "Investor Update" and body "Latest metrics attached" and return draftId, subject
Archive a Message: 
Archive emails that have the following `Subject` and return id, folder
Start with a single, specific action (for example, “Read unread emails from yesterday”). Once you see the desired output, chain additional steps with other nodes such as Ask AI or Google Sheets Writer for further processing.

​
Troubleshooting

If your Outlook MCP node isn’t working as expected, try these best practices:

​
Keep Prompts Simple and Specific

  • Good: “Read unread emails from jane@contoso.com and return subject, sender”
  • Bad: “Read unread emails, summarize them, and then send a summary to my manager”
For best results, create one Outlook MCP node to fetch the emails, then use separate nodes (for example, Ask AI and Send Email notification) to handle summarizing and sending.

​
Match What Outlook Can Do

  • Good: “Forward email from recipient to tim@acme.com and return forwardId”
  • Bad: “Forward the email and then post the summary to Slack”
Outlook MCP excels at email operations. To post to Slack, add a Slack Message Sender node after Outlook retrieves or forwards the message.

​
Break Complex Tasks Into Steps

Instead of trying to do everything in one prompt (which might cause timeouts or missed details): 
Read unread emails from last week, summarize each one, create drafts to reply, and then mark all as read
Break this into smaller, focused nodes:
1

Step 1: Read Emails

Read unread emails from 2025-03-10 to 2025-03-17 and return id, subject, bodyPreview
2

Step 2: Summarize Using `Ask AI` node

Summarize the email body and output the summary
3

Step 3: Draft Replies

Create a draft reply to email id id with body summary and return draftId
4

Step 4: Mark as Read

Mark email id id as read and return id, isRead
Connect these nodes sequentially in your workflow. The ids from Step 1 feed into Steps 2-4 so each action runs on the correct email.

​
Focus on Data Retrieval

Outlook MCP is great at getting information from your mailbox. For deeper analysis or content generation, connect it to other nodes. Example:
  • Good prompt: “Read emails flagged as important and return subject, sender”
  • Bad prompt: “Read important emails and generate a Slide deck summarizing them”
Use Outlook MCP to retrieve the data, then hand it off to Ask AI, & Google Slides Writer, for analysis or document creation.

​
Troubleshooting Node Creation

​
Need More Help?