- Engineers were pulled into ticket triage constantly, regardless of whether the issue actually needed their skill level
- Sprint commitments slipped because unplanned support interruptions were never accounted for in planning
- The support team could not self serve more without engineering context they simply did not have
- Ticket classification separated routine issues from ones that genuinely needed engineering judgment
- Context retrieval pulled relevant past tickets and documentation automatically into every routed case
- Support staff resolved a meaningfully larger share of tickets without escalating to engineering
The bottleneck was never engineering skill. It was engineering attention being spent on tickets that never should have reached them.
Hary Periya
- The irony is hard to miss. Companies building AI products were not using it on their own workflows
- Engineers did not need fewer tickets. They needed the right tickets reaching them
- The biggest gain here was not a new model. It was routing logic nobody had built yet
- Engineering interrupt volume from support tickets dropped meaningfully within the first month of use
- Sprint commitment accuracy improved once unplanned interruptions stopped derailing planned work
- Average ticket resolution time improved across both support and escalated engineering tickets
Engineering Interrupt Volume Reduced
0
%
Tickets Resolved Without Escalation Monthly
0
+