Our team is exploring Microsoft Semantic Kernel to see if we can build an agent that reads our Jira tickets and suggests a Sprint Goal. Since it's a Microsoft tool, I assume the integration with Azure DevOps or Jira is seamless? How would this affect our existing Agile and Scrum ceremonies?
3 answers
The integration is actually one of the strongest selling points for Agile teams. You can use the "OpenAPI" plugin feature in Microsoft Semantic Kernel to connect directly to the Jira or Azure DevOps REST APIs. We built a grooming assistant that analyzes the description of every ticket in the "Ready" column and flags any that are missing acceptance criteria. It doesn't replace the ceremony, but it makes the "Pre-Grooming" phase much faster. During the planning meeting, the agent can suggest a Sprint Goal based on the common themes of the top-priority tickets. It’s a great way to use AI to keep the team focused on high-value work without adding administrative overhead
How do you handle the permissions? You don't want the AI agent to have the power to delete tickets or change priorities by mistake, right?
It also helps with "Sprint Velocity" predictions. It can look at past performance data and warn if the team is over-committing.
Exactly, Lawrence. Using the historical data as "Semantic Memory" makes those predictions much more accurate than a simple average.
Great catch, Franklin. In Microsoft Semantic Kernel, you control exactly what functions the agent has access to through the "Kernel Plugin" definition. We create a "Read-Only" plugin for the grooming phase. The agent can suggest changes, but a human Scrum Master has to call the "Commit Changes" function. It keeps the AI in a supportive role rather than giving it total control over the backlog.