BONUS AI as the Third Perspective for Agile Delivery With Rohini Prava

BONUS: AI as the Third Perspective for Agile Delivery AI is already changing how agile delivery managers, Scrum Masters, and software teams handle the day-to-day work of delivery. In this BONUS episode, Rohini Prava shares how her team moved from experimenting with Cursor to using AI for practical work: bug analysis, ticket preparation, test planning, reporting, and reducing painful meetings. From Java Developer to Agile Delivery Manager "I see a good fit. And one of the scrum masters was going on holiday and I got the opportunity, I tried it, I liked it and I think I did well because I was given the role." Rohini started her career as a Java developer in India, then moved into people management and delivery-focused work before relocating to the UK. A temporary opportunity to cover for a Scrum Master opened the door to a new direction. That experience eventually led her into the Agile Delivery Manager role, where she now leads a distributed team delivering insurance solutions. When AI Became Part of the Work "It stopped being a curiosity when I integrated it in my work. And the fun started." For Rohini, the shift happened when her company gave access to Cursor, a developer-focused AI tool. Instead of treating it as something only developers could use, she tested it against real delivery work. The first meaningful example came before a retrospective-style meeting on a difficult project. By giving AI access to project data from Jira Service Management and Azure DevOps, she generated an eight-page analysis in under an hour. The result was a more factual, constructive conversation instead of a meeting based on memory, frustration, or blame. AI Across the Delivery Team "It really helps the devs to investigate issues, going through the legacy code, because some of the technology they may not even be familiar with." Rohini describes AI adoption as early, practical, and team-specific. Her team connected Cursor with Azure DevOps through the Model Context Protocol concept, giving the tool useful context about tickets and code. Developers use it to explore legacy code, brainstorm solutions, and support pair-programming-style thinking. Business analysts use it to draft tickets and acceptance criteria. QA team members use it to strengthen test plans and coverage. Rohini uses it for reporting, meeting preparation, email drafting, and pulling useful data from delivery events. Measuring Impact Beyond Speed "If you're just simply going with, oh, I've used it for quickly getting the reports, but we need to sit and look at, do we really need that report?" Rohini is careful not to frame AI only as a speed tool. Her point is that teams need to measure whether AI changes anything that matters. One personal example was timesheet review: a weekly task that used to take five to six hours. By building a small tool to help with review, she reduced that work to about one hour. The lesson for agile practitioners is direct: start by looking at the work that consumes time, then ask whether AI actually reduces waste or simply creates more output. Using AI to Reduce Meetings and Improve Quality "AI showed us that you can do less meetings, but just focus on one or two actions and see what change you can bring." One of the strongest team examples came from long root cause analysis meetings. The team had many bugs and spent 60 to 90 minutes with 11 or 12 people trying to analyze them. The meetings produced long action lists, but those lists were hard to turn into focused improvements. Rohini gave AI six months of bug data and asked for the most painful patterns. The suggestion was simple: QA slippage was high, and the team needed an "escape reason" field to understand why bugs were missed before reaching customers or UAT. That small change helped the team move from broad discussion to focused learning about test plan gaps and test coverage. The Risks: Over-Reliance, Overuse, and Governance "We don't have to automate everything just because we have AI. We don't have to include it in every task." Rohini warns teams not to trust confident AI answers without checking them. AI can produce well-structured responses that feel authoritative, but teams still need to validate the output against their own context. She also warns against overuse: AI does not need to be inserted into every task simply because the tool is available. For organizations, governance and security matter as well, especially when project data, customer context, source code, and delivery information are being exposed to AI-enabled tools. Start Small, Inspect, Iterate, Adapt "Start something basic, small that you know how to validate." Rohini's advice for Scrum Masters, Agile Coaches, and Agile Delivery Managers is to start with a familiar, time-consuming task. Pick something you understand well enough to evaluate. Ask AI to help with version one. Measure the result, validate it, then iterate after a week. This gives teams a practical way to build judgment before using AI on larger or more ambiguous problems. About Rohini Prava Rohini leads a diverse, distributed software development team delivering insurance solutions. She bridges business strategy and technical execution, managing stakeholders, mitigating delivery risks, and reducing cycle times. A composed, detail-oriented leader, she coaches autonomous, cross-functional teams, fostering innovation, continuous improvement, strong customer satisfaction, and meaningful impact across organizations. You can link with Rohini Prava on LinkedIn.