With AI agents increasingly becoming part of development teams, I see more people questioning whether Scrum is still relevant. If AI can complete work in hours that previously took days, does it still make sense to plan in two-week Sprints? Do we still need the same events and ways of working?
I actually think these questions expose a much bigger problem. Perhaps AI isn’t making Scrum irrelevant. Perhaps it’s exposing how badly many organisations have misunderstood Scrum in the first place.
Have Organisations Completely Misunderstood Scrum?
Scrum was created to help teams develop products in complex and uncertain environments. Environments where we don’t have all the answers upfront, where we need to experiment, learn from feedback and continuously adapt.
Yet many organisations have turned Scrum into a framework for micromanaging work. They expect detailed Sprint plans, estimates for every task and predictable delivery of everything committed to at the beginning of a Sprint. The focus becomes whether the team completed its tasks, rather than whether it achieved something valuable.
In many cases, organisations have simply taken traditional project management, divided the work into two-week periods and called it Scrum. And when that doesn’t work, Scrum gets blamed.
But if we look at what Scrum was actually designed for, isn’t it more relevant than ever in the age of AI?
We don’t yet know exactly how AI agents will change the way teams work. We don’t know how much faster they will make us, where new bottlenecks will appear or which practices will need to change. And that uncertainty is precisely why we need an empirical approach, where we experiment, inspect the results and adapt based on what we learn.
Are We Too Focused on Planning the Work Instead of Achieving the Goal?
During a our recent Meetup, Leandri and Casper from ICETech joined Bevan Williams and me to discuss their plans to introduce AI agents into their development teams. One of the questions raised was how to plan two weeks of work when you don’t yet know what AI agents are capable of delivering. It’s a very relevant question, but I think it also exposes how we’ve come to understand Sprint Planning.
A Sprint shouldn’t be about predicting every task the team will complete over the next two weeks. It should be about agreeing on a meaningful goal and determining how to work towards it, knowing that we will learn along the way. And AI makes this even more interesting.
Imagine that instead of giving an AI agent a detailed list of features to build, we gave a team a problem to solve. What if we could use several AI agents to explore different solutions, test them and learn which approach creates the most value?
During the discussion, I used the example of an organisation deciding it needs to build a car. It creates a business case, secures funding and starts planning all the work required to deliver the car. But what if the actual problem was simply getting people from A to B? By deciding on the solution upfront, we’ve already limited our ability to innovate.
AI creates opportunities to experiment with solutions that previously might have been too expensive or time-consuming to explore. But only if we stop treating the plan as more important than the problem we’re trying to solve.
Perhaps AI Is Giving Us an Opportunity to Finally Use Scrum as Intended
What excites me about AI isn’t simply that teams might become faster. It’s that we have an opportunity to rethink how we develop products.
For years, many organisations have been so focused on predictability, estimates and delivering against plans that they’ve lost sight of what agility is really about. Now AI is introducing a level of uncertainty that makes it increasingly difficult to pretend we can plan and predict everything upfront.
Instead of trying to create the perfect process for AI agents before we even understand how they will perform, why not start experimenting? Set a meaningful goal, try something, measure what happens and use that learning to decide what to do next.
That’s also what I appreciated about the discussion with ICETech. They weren’t claiming to have all the answers. They were preparing to experiment and wanted to understand what they should pay attention to as they learned.
Perhaps that’s the mindset more organisations need to embrace.
AI isn’t making Scrum irrelevant. It’s making the principles behind Scrum more relevant than ever. But only if organisations are willing to move beyond managing tasks and start focusing on goals, learning and delivering real value.
Watch the Conversation
This article was inspired by our Meetup, where Leandri and Casper from ICETech joined Bevan Williams and me to discuss the practical challenges of introducing AI agents into development teams.
Watch the full recording: When AI Joins the Team, Does Scrum Still Make Sense?
Further Reading
But this challenge goes beyond Scrum. Many organisations are now approaching AI transformation with the same thinking that prevented their Agile transformations from delivering the expected results.
Is Your Organisation Repeating the Same Mistakes With AI That It Made With Agile?
_____________
About the author: Malene Jacobsen is the founder of Agile Training Alliance, a Certified Scrum Trainer and a member of the Scrum Alliance Board of Directors. She has extensive experience helping organisations improve their ways of working, develop organisational agility and focus on delivering real business value.
