Every service desk has one: the shared mailbox. Support@, facilities@, purchasing@. Everyone can open it, so nobody feels they own the mail from 08:12. Someone reads it, assumes a colleague will pick it up and two days later the sender calls.
The work that follows is the same every time. Read it, understand what it is about, create an incident, file the mail. That is exactly the kind of work Flows takes over.
Why forwarding is not automation
The classic answer is a forwarding rule or a mail relay: the mailbox passes every message on to an address of the ticket system. That works, until it does not. The ticket gets the subject line as its title, even when that is "RE: RE: question". Nobody knows which mail became a ticket and which did not. And the rule lives in the settings of one mailbox, outside any overview.
A flow does it differently. It reads the mail itself, understands it and creates the incident with a summary a person would write. And it does so under rules that were approved in advance.
Start from the recipe
You do not have to design this yourself. A recipe is ready: turn emails into incidents. It does four things.
- Every five minutes it checks the shared mailbox for unread mail.
- An AI step takes a short summary, the details and the reporter from each mail.
- An incident is opened in ServiceNow with that summary as its title.
- The mail moves to a folder for handled messages.
If ServiceNow refuses the incident, the mail stays unread and the flow tries again. Nothing falls between the cracks.
A recipe ships with sample data, so the first test run completes before anything is connected. After that you fill in two things: which mailbox and which folder.
Or start the moment the mail arrives
Five minutes is fast enough for most teams. Where it is not, a flow can start as soon as Microsoft 365 reports new mail. Without a mail relay in between.
It helps to know how that works. The notification carries only the message's ID, not its content. The flow then fetches the message itself, through its own Microsoft 365 connection that an administrator approves first. That connection is separate from the notifications to Teams and Outlook: those only send.
One honest note belongs here. Microsoft lets such a subscription to new mail expire after about three days, and Datahub does not renew it for you. You build the flow that keeps it alive. If a flow hears nothing for three days, the administrator gets a notice. Silence gets noticed.
Who says yes beforehand
A flow that reads mail and creates incidents touches two systems. That is why there are two moments at which someone else looks along.
The connection. An administrator approves the connection to the mailbox and to ServiceNow, once. Whoever asked for it does not approve it themselves.
The flow. A test run does not carry out its writes but records what it would do. After that a colleague approves the version before it goes live. The flow runs with its owner's access and no more.
If something goes wrong later, an error flow reports it in the team's channel: which flow, which step and why.
Where to start
- Pick one shared mailbox where work now waits.
- Start the recipe and look at the test run on sample data.
- Have an administrator approve the connections.
- Let the flow run for a week and read along with what it creates.
The goal is not fewer people on the service desk. The goal is that nobody retypes mail any more and that every question has an owner within five minutes.
