The Agile Manifesto turns 23: is it time to retire?

Three major factors in software development are technology, ways of working, and, most importantly, people. These factors are deeply interconnected: people decide ways of working and technology, yet tools and technology, in turn, influence on ways of working. For example Github pull requests had a big impact on how teams develop software, review each other's work and release software. Ways of working have a major impact on how people feel when working which is reflected on job satisfaction. High job satisfaction makes recruitment and retention easier, creating a cycle where each factor depends on the others.

The Agile Manifesto was published in 2001 and it has had a huge impact on software development ever since. Among other things, it famously, and I think correctly, states that individuals and interactions should be valued more than processes and tools. Software is, after all, peopleware.

Two significant disruptions in recent years—COVID-19 and the rise of AI—have fundamentally altered the landscape. COVID made remote work and distributed teams reality everywhere and thus changed fundamentally the way people interact with each other. Although distributed teams had been gradually gaining popularity, COVID-19 accelerated the transition in a way no one could choose or avoid. Adaptation to the circumstances was the only option. The shift made interactions harder, especially for the teams which hadn’t worked remotely earlier. The impact was reflected for example in the DORA questionnaire where the average results measuring software team performance dropped. Even after COVID-19, many people stuck to working from home, for example it was felt to bring better work life balance. The impact of remote work to teamwork varied a lot: some teams were able to adapt to work very efficiently, others not as the interactions became harder and they were not able to replace face to face communication with equally efficient manners.

Why are "individuals and interactions over processes and tools" considered so critical in the Agile Manifesto? The reason is two folded. The first one is that there needs to be communication with the stakeholders giving feedback on digital product development which ensures that right things are being done. Another is that communication enhances productivity, doing things right and thus even occupational wellbeing. The problems get resolved faster when you have peer support and face to face communication requires less effort than making calls or writing messages.

Another major change has been the increase of usage and capabilities of AI. Generative AI has transformed software development with AI assisted tools offering the next level of code completion, pair programming buddy and an alternative for searching examples from Stack Overflow. As the personal interactions have come harder with the increase of remote work, the tools providing assistance have leaped on enormously. AI assisted coding tools like GitHub Copilot, and chatbots like ChatGPT offer capabilities which were utopian in 2001 when Agile Manifesto was written. Thus the tools nowadays have capabilities that another person was needed for 20 years ago. You can pair programs with AI, getting useful replies and tips for enhancing the quality of your code.

Thus there’s a question that is Agile Manifesto outdated? Has circumstances made communication harder, but have the tools caught up the difference and can they be your new peer? I would say yes and no. Yes in that sense that you have a tireless peer who answers your questions on a wide variety of topics. The quality of the responses may warie too and a human must be firmly on a driver’s seat. The AI tools put emphasis to individual performance by giving advice. They’re not working (yet?) at team level or helping in architecture or project management. Thus it can be said that they have a low abstraction level at the moment. However, that is definitely going to change as LLMs are getting more powerful. An AI which follows discussions, Slack chats and code changes in repository and gives proactively feedback based on all activities isn’t probably too far away.

As it stands currently, higher the abstraction level is, the work is human directed and the best results are achieved with co-operation in multidisciplinary teams where people get new ideas and can brainstorm together. AI will learn also in such an environment, but having an AI which would think out of a box is still far away.

In the end, it is important for the teams to leverage the advanced tools while holding on to agile principles to ensure that the right things are done in the right way.