Jira Issue Auto-Triage and Labeling

Overview

This flow speeds Jira issue triage with AI while keeping the team in control. When a new issue is created in a watched Jira project, the flow sends the summary and description to Azure OpenAI, which classifies it (component, severity, priority, suggested labels, suggested assignee, duplicate likelihood, confidence). It then searches Jira for likely duplicates, posts an AI-assisted triage comment on the issue, and - only when the model's confidence meets a configurable threshold - applies the recommended labels and priority, then posts a triage digest to Teams.

Why it matters: initial triage is a bottleneck. AI-assisted classification cuts time-to-assignment, while a confidence gate, a governed label set, and an explicit AI-assisted comment keep humans in the loop.

Ships Off (demo).

Use Case

A team with high Jira issue volume wants consistent, fast first-pass triage. Rather than a human reading every new ticket to set component/severity/labels and spot duplicates, the flow drafts that triage automatically and applies it when confident, leaving low-confidence calls and final approval to people.

Flow Architecture

Environment Variables

Connectors & Connections

Note — All connections are referenced as solution connection references; the flow is portable between environments as long as a connection is mapped at import time.

Customization Guide

Almost every realistic variant of this flow can be implemented by changing environment variable values. A few cases require small edits inside the flow definition — those are called out explicitly below.

Confidence gate

flowlibs_ConfidenceThreshold controls when labels/priority auto-apply. Raise it to be more conservative; the comment is always posted.

Auto-assign

Map components to accountIds in flowlibs_AssignRules and add assignee accountId to the applied fields to auto-assign (Jira Cloud needs real accountIds).

Duplicate linking

Extend the duplicate search results into a Jira issue link when duplicate likelihood is High.

Feedback loop

Log AI vs human-corrected triage to Dataverse to measure accuracy and tune the prompt.

Trigger cadence

OnNewIssue_V2 polls every 5 minutes; adjust or switch to a webhook pattern for lower latency.

Key Expressions

The flow is intentionally light on Power Fx / WDL gymnastics — the heaviest expressions are the branch-name concatenation and the approval outcome check. They are listed below in the order they appear in the flow.

body('Classify_Issue_With_Azure_OpenAI')?['choices'][0]?['message']?['content']

greaterOrEquals(float(coalesce(body('Parse_AI_Classification')?['confidence'], 0)), float(variables('varConfidenceThreshold')))

project = {key} AND statusCategory != Done AND key != {issueKey} AND text ~ "{summary}" ORDER BY created DESC

first(body('Search_Possible_Duplicates')?['issues'])?['key']