AI and Agile Coaching - Extension, Not Replacement
During Give Thanks for Scrum 2025 in Burlington, I was asked a question that seems to be coming up frequently: “Are you building something to replace Agile coaches?”
My answer was no, and will always be no.
That question led to deeper conversations about how Scrum Masters and Agile Coaches should think about AI, not as a substitute for coaching judgment, but as a way to reduce cognitive load, improve preparation, and reinforce good practices.
I cowrote a piece exploring this perspective: how AI can support coaching workflows, where it helps, and where human judgment remains essential.
Sharing here to contribute to the discussion and hear how others in this community are thinking about AI in their own coaching or Scrum practice.
Optional reading:
https://medium.com/@igharios/how-ai-can-enhance-productivity-in-agile-c…
I appreciate the care taken to frame AI as an extension of coaching judgment rather than a replacement. That said, I struggled with one core assumption in the article and this is where you lost me:
Agile coaches and scrum masters spend hours manually updating boards and tickets, consolidating metrics from multiple tools, preparing reports for stakeholders, and tracking down status updates that should be readily available.
The claim that Agile coaches and Scrum Masters do those things reflects a model of the role that Scrum is explicitly designed to eliminate. In healthy Scrum implementations, that work is unnecessary or intentionally avoided to surface deeper issues around accountability, transparency, and leadership behavior.
From a coaching perspective, the problem is not cognitive load that needs better tooling. It is system design that assigns coordination and reporting work to roles meant to challenge those patterns. Automating that work risks making misapplied Scrum more efficient rather than enabling organizational maturity.
This distinction matters in a Scrum context, and I would be interested in how you see AI supporting coaching without reinforcing role misuse or management-by-dashboard.
One additional concern is the visual framing. The image suggests Scrum Masters reviewing progress and assigning "tickets", which reinforces a task-management or project-coordination model of the the Scrum Master's accountabilities. In Scrum, Scrum Masters do not assign work or manage progress; those responsibilities sit with the Developers through self-management and the Sprint Goal.
The workflow is representative of many set-ups where people think they have implemented Scrum. A few Scrum words have been applied as a veneer over existing reporting and control structures. The associated workflows can be automated in all sorts of ways, but the result will never be Scrum outcomes.
Coaching is one of a Scrum Master's many stances. A good coach is good at wondering and at developing a human train of thought. Most of the complexity we deal with in the real world comes down to people. All human life is there: people's politics, hopes, vanities, and ambitions. If AI can help deal with that constructively, rather than adding to the slop, then it might get a job.