Customers wanted incidents resolved with minimum manual effort. I shaped the strategy to get there: connect the response in JSM, capture how incidents were resolved, and build the knowledge foundation for automation.
Role
Senior Product Manager
Period
May 2021 — Jul 2024
Focus
Automated resolution strategy · Jira Service Management
The goal: less manual effort
An incident occurs. Teams still have to find responders, coordinate the investigation, update stakeholders and document the resolution. Customers wanted that work reduced, but largely used JSM as a ticketing system while resolution happened elsewhere.
As Senior Product Manager, I owned the end-to-end Incident Management experience and shaped the strategy toward automated incident resolution. Bringing the work into JSM and capturing resolution knowledge were the foundations for that ambition.
1. Connect the response
The first step was to bring incident response into JSM. I built the capabilities that helped teams mobilise people, collaborate and communicate within the incident management experience.
Mobilise responders
Add responders and alert them to bring the right people into the response.
Coordinate the work
Build the Incident Command Center, integrating Slack and video conferencing for collaboration.
Keep stakeholders informed
Provide real-time updates while the incident is being handled.
Capture what worked
Introduce post-incident reports to preserve how the incident was resolved.
Across the broader incident management experience, monthly active users grew from 60,000 to 340,000 and CSAT improved from 43 to 72.
2. Build the knowledge foundation
The long-term strategy was automated incident resolution: a path from playbooks to runbooks and intelligent resolution. Learning from past incidents required knowledge of how they had been resolved.
But writing a post-incident report still meant more manual work. Major-incident closure and report creation stood at 0.2%. Introducing PIRs was a foundation; making them easier to complete was the next step.
Make knowledge easier to capture
I built AI-generated post-incident reports from incident context. Users received a draft to review and approve, rather than having to write the account from scratch.
01Gather context
Bring incident information together.
02Generate a draft
Turn that context into a report.
03Review and approve
Keep users in control of the record.
Atlassian was customer zero. Shadow testing compared generated summaries with customer-written reports, followed by customer testing and phased releases. Each phase had quality gates, with an accuracy threshold above 80% as a release criterion.
Progress toward the ambition
0.2%Starting point
14%1 quarter
34%3 quarters
Major-incident closure and PIR creation
The rate reached 14% in one quarter and 34% over three quarters. Easier reporting helped capture the knowledge needed for the longer-term goal of automated resolution.
MAU and CSAT reflect the broader incident management experience. Automated resolution was the strategic destination, not a delivered outcome claimed here.
60K → 340K
Incident management MAU · Dec 2021–Jun 2023
43 → 72
Customer satisfaction score
0.2% → 34%
Major-incident closure & PIR creation · 3 quarters