Before
Support teams monitored Slack conversations manually, read message threads, and switched to Notion to look up policies and supported request types. They then moved to Jira Service Management to create and categorize tickets, copied information between tools, and coordinated follow-up or approval by hand.
How it works
- Capture the Slack request, filter irrelevant events, and retrieve the full conversation thread.
- Use request types and store policies to classify the request, detect duplicates, identify the beneficiary, collect missing information, and determine whether approval is required.
- Create and categorize the service ticket, request or validate approval when needed, and post the outcome and ticket details back to the Slack thread.
Workflow details
- Request capture and context: The workflow starts when a Slack message event is received. It filters out retries, bot messages, edited messages, and deleted messages, then retrieves the relevant conversation thread so the full request context is available for processing.
- Policy and request validation: It loads supported request types and store policies from Notion before assessing the request. The AI agent checks for duplicate or already-active requests, recognizes informational messages, identifies the beneficiary, and determines whether required details such as the application, equipment, purpose, entity, urgency, or affected users are present.
- Clarification and approval handling: When information is missing, the agent asks for the needed details together in a single Slack response. For requests that require approval, it uses the available user profile information to identify the appropriate approver, posts the ticket and approval status in the thread, and validates that a response comes from the expected approver before continuing.
- Ticket creation and categorization: Once the request is sufficiently complete, the workflow creates a Jira Service Management ticket on behalf of the beneficiary and applies the relevant request type, affected service, priority, due date, and other policy-defined fields where applicable. Multiple-user or multi-application requests can be combined into one ticket when appropriate.
- Requester communication and summary: The requester receives a concise threaded Slack update containing the request status and ticket link, including approval or urgent-incident details when relevant. The workflow also produces a structured summary covering the classification, beneficiary, ticket, approval status, and any remaining blocker.
What changed
Slack requests are now processed through an automated sequence of context retrieval, duplicate detection, classification, validation, approval checking, ticket creation, and requester notification. Human involvement is limited to decisions or approvals that the policy requires, while routine ticket preparation and communication happen within the workflow.
Work removed
Manually reading Slack threads, checking request types and policies, identifying beneficiaries, asking repeated clarification questions, switching to Jira Service Management, copying request details into tickets, categorizing tickets, and manually posting status updates.
Expected outcome
Estimated 5–10 minutes saved per request, with more consistent ticket quality and fewer requests stalled while someone gathers context.
