.webp)
JSM in Slack and Teams: Set Up a Chat Help Desk
From the archive
How do you set up a JSM help desk in Slack or Teams?
Connect the service space to a chat platform through Atlassian Assist, choose the request types and channels, and test a complete request before rollout. JSM chat connects a service space to either Slack or Microsoft Teams; choose the platform for each space deliberately.
- Prepare intake. Choose the service owner, request types, required fields, and customer access rules.
- Connect Assist. In the space's Chat settings, connect the chosen platform and complete its administrator approval and onboarding steps. For Teams, follow Atlassian's Teams setup instructions.
- Separate channels. Configure the agent team and requester channels, with private paths for sensitive matters.
- Test the record. Submit a request, reply from both sides, and check the request type, comments, notifications, SLA, and visibility.
- Add AI selectively. If entitled, configure the virtual service agent with approved knowledge, escalation paths, and a usage limit. Measure a small pilot first.
Your IT team isn't running a service desk. They're running an unindexed DM queue that occasionally produces tickets. Same for HR. Same for Finance. Same, probably, for whichever poor soul in Legal keeps getting pinged about NDAs at 4:47 PM on Fridays.
The work is getting done. The data isn't getting captured. And the people doing the helping have no way to prove how much of their week is spent helping.
There's a fix, and it's been quietly maturing inside the Atlassian stack for years: run Jira Service Management directly inside Slack and Microsoft Teams. With the Team '26 Rovo updates layered on top, this isn't just possible — it's the way you should be thinking about every internal service team in your organization.
The portal problem nobody wants to admit
A portal launch can leave an old habit intact: people still DM whichever helper they know. Before changing intake, measure how much work arrives outside the service desk and which channels employees actually use.
People often ask for help where they already work. A chat entry point can capture that habit while the portal remains available for people who prefer a structured form. The design goal is consistent tracking, whichever supported channel they choose.
You can fight that. Or you can route around it.
What “JSM in Slack/Teams” actually does in 2026
The Atlassian-native integrations for Slack and Microsoft Teams have quietly become genuinely capable. The current state of play:
- Raise requests from inside a conversation. Slash command, message action, or emoji reaction. A guided form appears inline. No portal hop required.
- Turn any message into a ticket. Someone DMed the wrong person? Right-click the message, “Create request.” The conversation context travels with it.
- Approvals in chat. Managers approve software access, time off, hardware requests, and contract reviews — all from Slack or Teams without ever opening Jira.
- Two-way sync. Agent comments show up in chat. Requester replies update the ticket. The conversation and the ticket are the same artifact.
- Channel-as-service-desk. Run an active triage channel where your team chats and works publicly — and any message can become a tracked, SLA-bound request without breaking flow.
- Rovo virtual agent. This is the bit that changed everything in the last twelve months. The virtual agent answers from your knowledge base, gathers missing info conversationally, triages priority, and only escalates to a human when it can't resolve. Its answers depend on the knowledge and flows configured for that service. Atlassian recommends current, clear, well-structured knowledge; check answer quality against your own requests instead of assuming a universal improvement rate.
This isn't just for IT
The most common mistake we see: organizations buy JSM, deploy it for IT, and stop. Internal JSM is a service-team chassis, not an IT tool. The same Slack/Teams workflow runs equally well for:
- HR — onboarding checklists, time off, benefits questions, “where's the holiday calendar?”
- Finance — expense help, PO approvals, budget questions
- Legal — NDA requests, contract reviews, policy lookups
- Marketing and Creative — asset requests, brand approvals, design queue intake
- Facilities and Workplace — desk reservations, building access, equipment
For HR-specific request design, see our Jira Service Management for HR guide.
Every one of those teams is currently running an invisible queue out of DMs and channel pings. Every one of them deserves an SLA, a knowledge base, and a virtual agent doing the first round of triage.
The metrics that actually move
Track these four measures before and during the pilot:
Where a request needs a visual walkthrough, our Loom and JSM guide explains how to keep a recording alongside the service record.
- Time to first response. Measure acknowledgment time and time to a useful response separately; an automatic greeting does not mean the request is resolved.
- Volume of human-touched tickets. Track the share resolved without human intervention, the share escalated, and the share reopened. Establish your own baseline and review answer quality; no universal first-90-day resolution rate is established here.
- Hidden work, surfaced. Work that was happening in DMs now shows up in the data. Sometimes this is uncomfortable. It's also exactly what your leadership team needs to make staffing decisions that aren't guesses.
- Employee satisfaction. Compare satisfaction with resolution quality and wait time. Easier intake only helps if people receive a useful outcome.
The gotchas worth flagging before you commit
Three pitfalls we routinely steer clients around:
- Keep one system of record. Chat and the customer portal can both feed JSM. Capture the work once, keep the request types consistent, and prevent separate untracked DM queues. Agents work in JSM; the portal remains a requester-facing option.
- Train the virtual agent on what's actually true today. The knowledge base needs to reflect how the company actually operates right now — not the SOP that someone documented in 2021 and nobody has updated since. A virtual agent that confidently delivers stale answers is worse than no virtual agent. Audit and refresh the knowledge base before you flip the switch.
- Permissions vary by service team. Default visibility settings that work for IT will not work for HR or Legal. Channel-based intake is great for IT triage and a privacy disaster for benefits questions. Plan request-type permissions and notification scopes before you launch, not after the first complaint.
Why this matters more in 2026 than it did in 2024
The Team '26 announcements strengthened Atlassian's focus on AI-assisted service work. For this rollout, the concrete building blocks are Assist for captured chat requests, an approved knowledge base, configured conversation flows, and a clear escalation path. Broader agent integrations require their own access and action review; they are not a prerequisite for a useful chat help desk.
Translation: the chat-based service desk is no longer just a convenience. It's the surface where AI gets to be genuinely useful to your internal teams. The organizations that wire this up in 2026 are going to have a meaningfully different cost-to-serve curve than the ones still defending their portal a year from now.
This is the kind of project where the tooling is mature, the integrations are first-party, and the gating factor is design and rollout discipline. As an Atlassian Solution Partner, Avaratak Consulting helps internal service teams stand up Slack and Teams-based JSM with the virtual agent properly tuned, the request types properly scoped, and the change management properly handled — so the rollout actually gets used. If your team is one DM avalanche away from burnout, find us at avaratak.com.