30. 12. 2023
Atlassian, Service Management
11. 06. 2024
Davide Sbetti
Development, Events, NetEye
An XPerience: Discussing the Role of AI in the Agile World
Between the 4th and 7th of June this year, Bolzano had the opportunity to host XP 2024, the 25th edition of the premier international conference on Agile software development.
The scenario was NOI Techpark and, as Würth Phoenix, we were there and had the chance to participate in the conference with two talks about SecDevOps and the role of AI in the Agile world.
In this post, I’d like to discuss a bit our thoughts on the second topic so… what role will AI have in the Agile methodology? Will it potentially replace some/most of the roles currently established in the Agile world?
AI approaches, AI approaches everywhere
Let’s be frank, these days we see an ever increasing trend of AI approaches in a wide variety of fields, with the term AI nowadays used (and sometimes even abused) more and more.
Being the “topic of the moment”, we decided to imagine and discuss with the audience at XP where AI approaches might find a place in the Agile methodology, while trying to be… creative! Let’s look at a summary of some of the ideas that were discussed!
Can we go beyond a sprint chart?
AI techniques are usually better with (and in fact generally require) a large amount of data…. So what type of data do we have available when we want to apply an Agile approach to software development?
Well, a first idea could be to apply various techniques to analyze data related to our sprints, i.e., the short development cycles residing at the core of the agile methodology, which consist of a set of small issues (or stories) that should be accomplished in a certain (small) amount of time.
For example, the characteristics of the various issues (requirements, estimated effort in story points) could be the data on which we could base an analysis of how sprints went, including not only the already existing sprint charts and the common characteristics of most of the tools used in the fields, but with a focus on extracting insights that may not be immediately visible from a classic chart.
An interesting question might be why was the development pace faster (or slower) in certain sprints? Answering such a question, for example by employing various data analysis techniques, may allow development teams to find a balance in their sprints that’s also useful to the management in having more predictable results.
A lot of text
Another interesting aspect is the large amount of text that may be collected by a development team or manager when it comes to feedback or customer interviews. One possibility that may be considered, especially given the recent developments and results obtained by large language models, could be to apply such models to the results of customer, consultant or stakeholder interviews, leading to a sort of automated requirements analysis that may be seen as a helpful starting point to plan future work and issues.
Estimations proposals
Similar to the idea of analyzing the data about sprints that we outlined above, we could decide to analyze the data about the estimations a team gave to issues in the past when planning its work, clearly taking into consideration various characteristics of the issues. This may lead to a proposal of estimations for future issues that, while probably not being completely accurate, may still be useful to managers to obtain a quick overview of which development steps could be next.
And then another question arises: could these approaches ever lead to an automated sprint proposal?
Hey, are we forgetting about human relationships?
Okay, actually the proposals outlined above, while in certain cases quite realistic, were proposed as a provocation to the audience during the XP conference.
That’s because we strongly believe in the principles outlined in the Agile Manifesto and, in particular, in the importance of people and interactions over processes and tools.
What we see, at least at the current stage, is the role of AI techniques as tools, which for sure can and will simplify and speed up certain tasks, but that cannot substitute (at least at the current stage) the human interactions that are core to a successful development and that, in any case, will need human supervision as a key to success.
These interactions are not only important when it comes to customers and stakeholders really understanding the user stories leading to proper requirements, but also when we think of the development team itself, where different members participate with different personalities leading to different ways to approach certain tasks. These are all factors that we would like to take into account when it comes to planning the work that will be carried out in the very near future and is something that currently may be quite difficult for a model to perceive.
Conclusions
At XP 2024, we had the opportunity to discuss how AI approaches may influence and impact the way we apply Agile methodology.
We explored some possible applications of AI and data analysis techniques, brainstorming how they could be applied to speeding things up and simplifying certain aspects related to the application of an Agile methodology.
However, we also underlined how human interactions are still key for us when it comes to software development and Agile, with the AI approaches available at the moment which hardly seem able to capture all the possible nuances we need to take into account when we’re planning work for a team, made up by people with possibly quite different personalities.
These Solutions are Engineered by Humans
Did you find this article interesting? Does it match your skill set? Programming is at the heart of how we develop customized solutions. In fact, we’re currently hiring for roles just like this and others here at Würth Phoenix.
Davide Sbetti
Hi! I'm Davide and I'm a Software Developer with the R&D Team in the "IT System & Service Management Solutions" group here at Würth Phoenix. IT has been a passion for me ever since I was a child, and so the direction of my studies was...never in any doubt! Lately, my interests have focused in particular on data science techniques and the training of machine learning models.
Author
