Why Agile Still Matters In An AI-First World
Stephen Clark, Senior Program Manager at The Vanguard Group, Inc.
gettyRecently, I saw a post on X that read, “Remember scrum masters? Wild times.” It made me laugh. It also made me think about how quickly the conversation around software development has changed. Are scrum masters really becoming extinct?
I understand the frustration with the role. I’ve seen Scrum implemented poorly, where daily standups become status meetings, retrospectives become calendar obligations and teams measure success by how many tickets they close. I’ve also seen teams complete sprint after sprint while delivering very little meaningful business value.
I’ve experienced this from several perspectives. Before I started leading engineering teams, I worked as an Agile coach. Along the way, I’ve worked with plenty of scrum masters and seen teams operate with dramatically different levels of effectiveness.
That experience has shaped my view of where Scrum belongs. As roles continue to evolve, the discipline behind Agile delivery is becoming more important. AI is a big reason why.
AI has made software development dramatically faster. Engineers can generate code, write tests, troubleshoot defects and prototype ideas in a fraction of the time it once took. Product managers can explore concepts and create working prototypes with ease.
It also creates a problem that technology leaders should take seriously. When it becomes easier to build something, organizations become capable of building far more things. That doesn’t mean those things are worth building, though.
I’ve seen engineers and product managers get excited about a new AI capability and quickly move toward building something with it. The technology is interesting, and the prototype looks impressive. Everyone can see the possibilities. The harder questions come afterward. Who needs this? What problem does it solve? What value does it create? Will anyone actually use it?
AI can make those questions easier to overlook because the cost of experimentation has dropped so dramatically. That’s where I think Agile still has an important role.
Good Agile practices create structure around an inherently chaotic process. Software development involves changing requirements, competing priorities, technical constraints, organizational dependencies and new information arriving every day. A framework such as Scrum gives teams a way to navigate that uncertainty while continually reassessing what deserves their attention.
At its best, Scrum creates a healthy tension between what a team could build and what it should build. That distinction matters even more when AI increases our development capacity. Imagine a team that can now accomplish twice as much development work because of AI. If its priorities are poor, it can also produce twice as much low-value work. Velocity without direction isn’t progress.
A good scrum master should be helping the team ask better questions, remove obstacles and maintain focus. They should be willing to challenge a priority when the work no longer makes sense.
The technology industry has spent years talking about technical debt. AI introduces a new dimension to the problem because it can accelerate development faster than organizations can validate and support what they create.
We can now generate applications, integrations and automations with remarkably little effort. That means we can also generate applications, integrations and automations that nobody uses. I worry about that almost as much as I worry about poorly written AI-generated code.
An application that never reaches production may not create traditional technical debt, but it still consumes engineering capacity, cloud resources, support attention and organizational focus. Multiply that across hundreds of teams experimenting with AI, and the cost becomes significant.
Some experimentation is exactly what organizations should encourage. Engineers need space to explore new technology, and teams will discover valuable ideas by building things that don’t work.
The key is knowing when to stop experimenting and when to invest.
Agile provides a framework for making that decision. Teams can test assumptions, gather feedback and decide whether an idea deserves another iteration or should be left behind.
That discipline becomes increasingly valuable as the cost of building software approaches zero.
I don’t see AI and Agile as competing approaches. AI can help us work more efficiently. Agile helps us organize that work and continually evaluate whether it is producing value. These capabilities complement each other.
AI might help a developer solve a problem in an afternoon that previously took several days. Scrum can help ensure that the team is spending that afternoon on a problem worth solving.
AI can help turn an idea into a prototype quickly. Agile can help the team determine whether that prototype deserves to become a product. AI can help us execute faster. Agile gives us a mechanism to pause, learn and adjust.
I don’t think scrum masters are extinct. Some of the traditional work associated with the role will undoubtedly change. AI can summarize meetings, identify action items, surface dependencies and analyze delivery data. Those are improvements worth embracing.
The opportunity is to spend less time administering Scrum and more time practicing it well. The best scrum masters I’ve worked with have never been calendar managers. They’ve been coaches, facilitators and sometimes the person willing to ask the uncomfortable question that everyone else was avoiding.
That role becomes even more valuable when teams can build faster than ever. The technology industry has spent decades trying to improve how quickly we can turn ideas into software. Our next challenge is deciding what deserves that speed.
The organizations that get the most from AI won’t necessarily be the ones that build the most applications or generate the most code. They’ll be the ones that can identify valuable problems, experiment quickly, learn from the results and confidently stop pursuing ideas that do not create value.
AI gives us incredible tools for building. Agile gives us a discipline for deciding where to focus those tools. I still think we need both.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
