I recently recorded a podcast discussing AI and the future of knowledge work. The podcast will be out in a few weeks. The discussion shifted to the relevance of Scrum when AI becomes an integral part of the team. The backdrop to this discussion is the staffing layoffs at many technology companies, often justified by the claim that AI can perform the same work with fewer people. Surely, the debate went, a single person can deliver the whole increment without the need for a team, and thus remove the need for Scrum, which enables teams to work together to solve complex problems.
There are two parts to that discussion that we touched on: Will AI render teams obsolete, and is Scrum still relevant when you don't have a team?
Will AI render teams obsolete?
First, we need to establish a timeframe for that question. At some point, AI, robots, and technology may render humans obsolete for all tasks. Think of the 2008 Pixar film WALL-E. But in the next five years, will AI make teams obsolete?
It reminds me of my first experiences out of college, when I was exposed to corporate IT. In college, I worked either alone or with one other person and built huge systems to deliver on an assignment. Many thousands of lines of code. I wrote all the documentation and even did the presentation. It seemed odd to see a team of ten people doing the same amount of work, especially since the team had specialists for analysis, testing, and even design. Of course, I settled into this model and soon came to appreciate that work was much more than getting stuff done. And instead of spending time solving problems, I spent that time in meetings discussing how the work would be allocated to the teams.
Of course, the agile community railed against this model, encouraging organizations to build small cross-functional teams and allowing those teams to deliver results. However, as I visit large organizations that use Scrum, I see a hybrid model. Yes, Scrum Teams are now cross-functional, but organizations still comprise a combination of specialized roles and teams. Thus, work still requires the involvement of multiple teams, and there is overhead and complexity associated with delivering an increment. Technical debt and complexity persist, compelling organizations to involve more personnel and structure their operations accordingly to address these challenges. The Agile Product Operating Model would encourage a better approach, but that would require change, and change is rigid, slow, and requires motivation, such as a competitor or massive market disruption. Change for the sake of it is rarely successful.
Teams might be smaller, but teams are not going away anytime soon. AI will be a team member, a team member that always says yes and often gets things wrong! This means you need an approach that enables the team to work together and supports frequent inspection. And when I say work together, I mean inclusive of AI. AI will also require more inspection, as results are never assured. That will no doubt increase the importance of the Daily Scrum as the team discusses whether the AI should accomplish this task or that task, or if someone else should partner with them. The interesting thing about AI as a teammate is that it will require teams to lean into Scrum even more. New AI teammates are like new junior team members that benefit from the discipline and transparency of Scrum. As we learn what works or doesn’t work for AI, we will adapt our working practices, and because AI is itself evolving, we can not take anything for granted. Things that the AI tool you are using will do a poor job of one week will change with the release of a new version. Retrospectives will allow us to openly discuss how AI was integrated into the team. Unlike traditional human teammates, AI does not get hurt feelings or avoid talking about difficult experiences during the Sprint.
Individuals can still benefit from Scrum.
In most organizations, we will continue to have teams, but with the power of AI, there will be opportunities for individuals to build things on their own. In particular, in the area of rapid prototyping or general experimentation. Is Scrum still relevant for these projects?
There has been much written about the value of Scrum when working alone or in small teams. In a nutshell, this work describes the benefits of:
- Discipline - Scrum provides a structured way for an individual to approach a problem, discouraging them from getting sidetracked. The structure of a Sprint enables individuals to focus on their tasks. The concepts of a Sprint Goal and Product Goal allow the individual to step back and reflect on the work. The Product and Sprint Backlogs provide them with a to-do list to work from.
- Transparency - Unless you are building something for yourself, typically someone is paying for the work you are doing, and other people who will benefit from the product you are creating. Scrum is ultimately a tool for enabling people to collaborate and solve complex problems. Some of those people are stakeholders. By using Scrum, you create an environment that allows those people to see what you are doing and provide feedback.
- Empiricism - At the heart of Scrum is the idea that you do not know everything, you have to deliver frequently, and learn from that process. When working alone, there is often the feeling that you need to complete everything and then share it. I know I am guilty of this when writing a book or building a new class. Instead, Scrum encourages you to focus on validating assumptions, mitigating risks, and exposing value frequently. This can help to ensure you are on the right track.
Therefore, Scrum can still be beneficial even when working alone, and because of the nature of AI, the need for frequent inspection and adaptation in a disciplined way becomes more important. I do imagine that Sprint lengths will decrease, or Sprint Goals will grow as we learn how to do more with our AI tools. The biggest challenge will not be the Scrum Team (one person plus AI co-pilots), but getting access to other Stakeholders to review the Increment and help shape the direction of the work. Often, the most significant determinant of Sprint length is not capacity but the Sprint Review. Getting the right people to be there is always a challenge! AI will just make this even more of a challenge.
Fundamentals of Scrum are even more relevant today.
Scrum is almost thirty years old, and over the past thirty years, many people have questioned its relevance, value, and approach. Those questions have not invalidated the use of Scrum but improved its application and how Scrum is described in the Scrum Guide and supporting materials. The fundamentals of Scrum describe how most people should work when faced with a complex problem that requires a mindful, disciplined approach to its resolution. Professional Scrum adds to Scrum, providing an additional set of ideas including empiricism, the Scrum Values, and focusing on outcomes, which help you avoid merely following the Scrum process and checking boxes.
AI does not change this very fundamental approach. It enables Scrum Teams to be more flexible and self-contained by leveraging the power of AI as a team member or co-pilot. It also adds the ability to test a hypothesis rapidly by simulating users or customers. AI might be a breakthrough technology, but it is also a natural evolution of tools supporting Scrum Teams, making Scrum more relevant.
This realization motivated the creation of a new self-paced class, Professional Scrum Fundamentals. The class is designed for the next generation of Professional Scrum practitioners who can leverage Scrum to solve complex problems and deliver value to their stakeholders. The class is self-paced, allowing learners to progress at their own pace and ultimately demonstrate their knowledge, as it includes an attempt at the PSM I exam.
AI is not a threat; it is an opportunity.
AI is a set of tools that will help Scrum Teams deliver on the promise of Scrum by enabling them to deliver value more frequently. It could also reduce the size and number of teams, which will improve the flow of value and ultimately benefit Scrum. However, to realize the opportunities that AI offers, teams cannot forget the fundamentals of Professional Scrum, focusing on discipline, transparency, and empiricism. Those fundamentals provide a firm foundation, ensuring that teams not only deliver value but also continually improve their delivery. The irony is that AI is often portrayed as a threat to Scrum, when it may be its hero.