Use case · Documents and finance
Email attachments classified, renamed and filed to SharePoint — with a log in Excel
A workflow triggers on each new email, asks an agent to classify the attachment from its content and the email context, renames it to your convention (date, sender, type, reference), saves it to the folder that type belongs in and appends a row to a log sheet. Unknown types go to an "Inbox" folder with a Teams mention. It is the starter workflow: small, visible, and the base for the document flows that follow.
- Who it is for
- Any team with a shared mailbox that collects attachments — accounts, HR, procurement, legal, a client-facing inbox at an agency — and a folder structure someone maintains by hand.
- Systems involved
- Outlook / Microsoft 365 or GmailSharePoint / OneDrive / Google DriveExcel 365 / Google SheetsTeams notification
How the run goes
5 steps · logged per run
- Trigger01
New email with attachment
Shared mailbox
- Agent02
Classify and name
Type, counterparty, date, reference
- System03
File to folder
SharePoint / OneDrive / Drive
- System04
Log row
Excel 365 / Sheets
- Person05
Handle unknowns
Teams mention on the Inbox folder
How it runs, step by step
- 1
The trigger watches a shared mailbox (Microsoft 365 or Google Workspace). Each email with attachments starts one run; inline images and signatures are ignored by size and type rules.
- 2
An agent step receives the attachment text (via a quick extraction) plus subject, sender and body, and returns a type from your list — invoice, purchase order, contract, statement, CV, other — with the counterparty, a date and a reference it found.
- 3
The file is renamed to the convention (for example 2026-10-19_AcmeCorp_Invoice_INV-4471.pdf) and saved to the folder mapped to the type and counterparty. Folders are created when missing.
- 4
A row is appended to the log: date received, sender, type, filename, link, run ID. Types below the confidence threshold are saved to an Inbox folder and a Teams message tags the owner.
- 5
Once this runs for a week, the invoice type usually gets its own IntelliDocs flow — that is how most teams grow into document automation.
Limits — what this does not do
- Classification is by content and context; a scanned image with no text is classified from the email alone and usually lands in Inbox.
- Workflow runs are the metered unit: one email, one run.
- Not a records-management system; retention and permissions stay with SharePoint or Drive.
Questions about this process
The first pages' text and the email context — enough to classify. Full extraction is a separate step with IntelliDocs when you need the fields.
Yes; each is classified and filed on its own, and the log gets one row per file.
It is the usual first sandbox process: live within a day, visible to the whole team, and the base for the next one.
Related use cases
Build this on your systems in a guided sandbox.
Tell us the process; after a 30-minute call the sandbox is ready from 24 hours, and we build the first run with your team.