As of
How to get employees to stop DMing IT in Slack and still track the work
You don't need employees to stop DMing IT in Slack; you need every DM to become trackable, and a DM goes untracked because it arrives as a display name and a sentence that someone has to identify before they can act. STLabs answers DMs with an AI agent that already knows who is asking from Axiom™, logs what it does against that person, opens a ticket with context when a person is needed, and shows employees their requests in My Requests in Slack.
Key takeaways
- A DM to IT goes untracked because it arrives as a display name and a sentence, and working out who is asking takes longer than answering.
- Request channels and forms make work trackable by asking the employee to supply what IT should already know, which is why people drift back to DMs.
- A request channel with Slack's built-in Workflow Builder forms gives requests a shape, but not an owner, a status the employee can check or a report.
- An AI agent that already knows who is messaging, their role, devices and access, can treat a DM as a request without asking a single form question.
- In STLabs, requests that need a person become tickets with identity, device, access and recent changes attached, and employees check status in My Requests in Slack.
Work moved into chat years ago, and IT requests followed. Most service desks have answered the same way: point people at the portal, then at a request channel, then at a form inside the channel. Each step is reasonable. A queue gives work an owner and a ticket gives it a history, and none of that exists in a DM.
The DMs keep coming anyway. The usual explanation is that people message where they work, so meet them in chat. That's true, and every chat tool says it. It doesn't explain why a DM in Slack is untrackable while a message in a request channel, in the same Slack, mostly is.
The difference is what has to happen before anyone can act.
Why are DMs to IT so hard to track?
Because a DM arrives as a display name and a sentence, and nothing else. "Can you get me into the finance dashboard?" says nothing about which finance dashboard, what the sender already has, whether their role allows it, or who would approve it.
So the person who receives it starts with a lookup. They open the identity provider to find the account, the HR system to check the role and manager, maybe device management to see the laptop. Only then do they know whether the ask is allowed and what doing it involves.
That lookup is the real work in most DMs, and it's done in someone's head and three browser tabs. Nothing about it lands in a queue. When it's finished the fix takes a minute, and logging it afterwards feels like paperwork for something already done.
It's also why DMs go to whoever answers fastest. An employee who messages the IT person they met at onboarding, rather than IT, is betting that this person already knows who they are. Often they're right, which is exactly the knowledge that never gets written down.
Why forms and request channels don't stick
Forms fix logging by moving the lookup onto the employee. Which application? Which device? Who is your manager? What access do you have now? Every field is something IT's own systems already hold.
That's the trade most service management setups make. The record is clean because the employee typed it in. The cost is that asking IT now takes longer than asking a person who knows you, and the DM wins again.
The display name causes quieter trouble too. Two people called Alex Kim, one in finance and one in engineering, can both show up as "Alex" in a busy thread. An analyst who grants the dashboard to the wrong Alex hasn't made a typing error. They've made an identity error that no form field would have caught.
Start with a request channel and Slack's own forms
None of this means the basics are wrong, and you can put them in place within a week.
Create one channel for requests, such as #it-help, and make it the official place to ask. Pin a note on what belongs there and add it to the channels new hires join. When a DM arrives, answer it if it's quick, then link the channel.
Then add a form with Slack's Workflow Builder: a shortcut that asks three or four fields and posts a structured message the team can work from. Microsoft Teams supports the same pattern with its own forms and workflow features.
For a small team this may be enough. But a message with a checkmark has no owner, no status the employee can look up and no monthly report, and the lookup still happens off the record.
When the channel isn't enough: chat help desks, service desks and AI agents
Once requests outgrow a channel, the tools fall into three categories.
Help desk apps built for chat. These turn a Slack or Teams message into a ticket with a reaction or a shortcut and add a light queue with assignees. They often fit a small team without a service desk. The ticket is a copy of what the employee typed, so an analyst still has to look up who is asking.
Service desks with a Slack or Teams integration. If you already run a service desk, its chat integration puts requests in the real queue, with routing, reporting and audit history. That's the better fit when an auditor reads your tickets. Many integrations answer a message with a form or a link to the portal, which puts the lookup back on the employee.
AI agents that resolve requests in the thread. These answer questions and carry out approved actions in the conversation. Whether they solve the DM problem depends on one thing: does the agent know who is asking, or does it ask? An agent that replies "What's your employee ID?" is a form with better manners. If most of your work is a person swapping hardware at a desk, a chat help desk app will serve you as well and cost less.
How an AI agent that knows who is asking makes a DM trackable
The DM stops being untrackable when the lookup is already done before the message arrives. If the agent receiving "Can you get me into the finance dashboard?" already knows the sender's role, manager, devices and current access, it can treat the message as a request: decide whether policy allows it, route the approval, do the work, and write it all down against the right person.
Nothing is asked of the employee that a form would have asked. They keep messaging the way they already do, and the record appears anyway.
Whether that record should also be a ticket is a separate question. The guide on whether every service request should be a ticket covers it.
How STLabs answers DMs and keeps the record
Employees ask STLabs in Slack, Microsoft Teams or the web portal, in plain language, in a channel or in a direct message to the STLabs app. Behind the agent is Axiom™, the live model of your organization that STLabs builds from your HR, identity provider and device management tools. Axiom™ keeps one record per person with every ID attached, so the Slack account that sent the message resolves to a person with a role, a manager, devices and access.
That's what settles the two Alexes. The agent answers the one in finance as the one in finance, using only what that employee is allowed to see, and draws on your wikis, knowledge base, past tickets and resolutions for their role and device.
For a request such as dashboard access, the agent checks the rules your team wrote in Policies & Approvals. A rule can allow a request, block it or require an approval, and the workflow waits for the approver the rule names. Because Axiom™ keeps role and manager current from HR, that approval goes to the current manager. Every AI decision and workflow run is logged in a searchable record, so every DM it handles leaves a record against the person who sent it.
When the agent can't resolve a request, it hands it to your team as a ticket with the requester's identity, device, access and recent changes already attached. The analyst starts where the lookup would have ended.
Employees can ask "What's happening with my dashboard request?" and see the current status, or open My Requests in the STLabs Slack app to see everything they've asked for. See Employee Self-Service.
This is the part an AI-native design changes: the agent's context is a live model of the organization, so a DM carries enough to act on.
How to test this on your own systems
Ask your service desk to tally every DM and side-channel request for two weeks: who asked, what about, and how long it took to work out who they were and what they had before anyone could act.
Take ten of those messages and send them, word for word, from the same employees' accounts, to the tool you're evaluating. Count how many follow-up questions it asks that your identity provider, HR system or device management could have answered. Check that each request is recorded against the right person, including any pair of employees who share a display name. Then check whether the employee can see where it stands without asking anyone.
Run the DM tally again a month after launch. If the DMs haven't dropped, the official route still asks for more than a DM does.
Let them message
You don't need to replace the service desk you already run. STLabs works alongside it, takes on one request type at a time, and teams go from integration to results in five days. The plan is in Moving to STLabs from your current service desk.
A DM was never the problem; a stranger's name on it was.
Frequently asked questions
Should IT ban Slack DMs?
No. A DM is untracked because it arrives as a name and a sentence, not because it's a DM. Answer it with something that already knows who is asking, and it can be tracked like any request.
Is a Slack channel enough to track IT requests?
For a small team it's a start. A channel with Slack's Workflow Builder forms gives requests a shape, but not an owner, a status the employee can check, or reporting.
How do ITSM tools track requests that start in Slack or Teams?
Most turn the message into a ticket, often after a form. An AI agent that knows the sender's role, devices and access can handle the message directly and log it against that person.
Can employees check request status in Slack?
Yes, if the tool supports it. In STLabs, employees ask the agent in plain English or open My Requests in the STLabs Slack app to see every request and where it stands.