The Impact of AI on Agile Product Development (until 2030)

The integration of AI in Agile product development will change the way agile teams work, make decisions, and deliver value to customers. By 2030, we can expect changes in various aspects of agility, from customer and stakeholder engagement to leadership, teams, and roles. Let’s go!

Three Futures

Let’s begin from the big picture, where I see three distinct futures:

1. The AI-Takeover is a dystopian future where AI systems replace human workers. AI managers are commonplace, task-mastering over humans. AI development and adoption boom. Super-massive efficiency wins left-right-and-center lead to super-massive unemployment. Ethical concerns. Political instability. Social unrest. Classic agile won’t work in this future because there are very few people around, in specialized narrow solo-roles.
2. The Human-AI Bromance is a symbiotic future where humans and AI systems work collaboratively, both leveraging each other’s strengths. Workforce becomes “AI literate” and adapts working models for customer value. We find pockets of increased productivity while maintaining at least some grounding in human-centricity. Potential for humans to focus on creative, strategic and empathetic tasks. Some layoffs, but also new blue oceans. Classic agile might not work in this future either, because teams as we know them today don’t exist anymore.
3. The “AI Bubble” Bursts is a failed paradigm shift where AI’s potential falters or crumbles. AI companies lose funding. AI-development and adoption stagnates. Supermassive regulation. Missed deadlines or production failures due to miscommunication or misinformation from an AI. Safety concerns. Loss of public trust in AI. Reversal back to human-centric work practices where classic agile process still works.

So, which future will materialize? I’m looking this as an investor, wanting to have a stake somewhere in expectation of future ROI, whilst minimizing my downside risk. With this mindset, here’s my %-prediction:

The winning future, or where I’d put my money is the 2nd future at a coin-flip chance of 50%. I think we can do some fantastic work in this future! 😁

Here’s the kicker; the 3rd future is only fractionally less likely at 49% chance. Why? Well, AI has had its run already. Technology evolves in leaps and I’m not at all convinced that AI’s pace could be on like Donkey Kong much longer. We might very well be stuck at this level of clever chat-botting and text-wrangling gymnastics for the next 10 years or longer. My gut is telling it’s time to call tops on the “AI bubble”. Yeah, call me “bubble boy”.

Lastly, the 1st future has a meek 1% chance of happening. Sure, the big AI guys promise us the moon, but let’s remember they are salesmen whose success hangs on us becoming believers of their “transhumanist fever-dream”. I’m too old to believe in fairy-tales.

Human-AI Bromance and Agility

Setting the focus now on the 2nd future, or human-AI bromance, for the remainder of this article. Here are the changes I’m expecting to see, per category:

Customers, Stakeholders, Marketing

1. Individually crafted customer value: The focus shifts to customer-problem-centric approaches, where AI helps identify and address specific customer needs. Long-lasting product-lines may become an obsolete strategy.
2. Stakeholder adaptability: Stakeholders and marketing must be prepared for rapid priority changes, and their feedback must be collected quickly in fast-paced environments. The risk with stakeholders and marketing is that they may end up becoming the “new bottleneck”.

Leadership

1. Balance: Leaders need to balance human needs to AI-capabilities. This change needs to be facilitated carefully because people are surprising beings. This may be the biggest challenge in management history.
2. Organizational agility: Companies will need to adapt their structures to facilitate faster decision-making, leveraging AI-driven recommendations and KPIs.
3. AI governance: Governance and oversight of AI will become a shared responsibility between leaders and stakeholders, ensuring that AI systems are aligned with business objectives and values.

Agile Events / Artefacts

1. Short sprints/iterations: Sprints may be reduced to just a few days, allowing for more rapid iteration and adaptation. However, I don’t think all teams shorten their sprints as downsides exist; for example the basic human need of mental reflection/celebration-time between completing a task and starting a new one, or “the tyranny of the increment”.
2. Streamlined meetings: The number of meetings and rituals will decrease. Events are mostly moderated by humans to ensure effective communication, collaboration and psychological safety.
3. Auto-generated/checked artefacts: The low-hanging fruit is to have some of the process artefacts and documentation auto-generated or auto-checked by an AI Agent. This fits well with the agile value “working software over comprehensive documentation”.

Teams

1. AI literacy: Team members will need to develop AI literacy to effectively collaborate with AI systems, leverage their capabilities and sanity-check their output.
2. Team composition: Smaller sizes (pairs, trios, quartets) and a mix of human and AI members. Adding AI-members can potentially break Brook’s Law ( “Adding manpower to a late software project makes it later”).
3. Custom team-processes: Tailored processes to a team’s needs and goals. As the need for snap-coordination increases, we’ll see more teams behaving like an open team or a super-group that live in a mutually beneficial relationships with the periphery. Their core is autonomic, interface open and mediated via agile interactions.
4. Purposeful team-life: Teams are more purpose-driven than ever before. A small team may directly serve an account (customer) for their immediate need. The same team can switch between accounts in rapid progression. This shortens life-cycles and puts focus on reteaming. We need to find ways to launch teams quickly (strategic speed) but mindfully (human soul’s slowness).

Agile Roles

1. Product Owner 2.0: The Product Owner role evolves to continuously adapt roadmaps, navigate customer-specific problem spaces, prime AI with context, and ensure unbiased AI use.
2. Scrum Master 2.0: The Scrum Master role transforms into a “catalytic counselor” with solid soft-, generic process- and AI-skills; understanding human needs and AI capabilities, drive change, and coach teams to become self-sufficient. No team should be left to fly completely alone in the dark.
3. Developers 2.0: Developer-sections, or cross-functional experts, will comprise humans and AI members. Humans are either true domain experts or prompt engineers who are guiding and validating a domain-expert AI’s work. The human members will have end-to-end responsibility for quality and KPIs, and will need to continuously upskill themselves in AI models, result evaluation, ethics, and more.
4. Shadow-Roles: A Team-AI can take obviously take various roles, including classical coder and tester. It can also take on shadow-roles such as shadow-customer (role-playing the user for feedback), shadow-product owner (predicting market needs) or shadow-scrum master (updating the sprint backlog and predicting impediments).

Agile Methods / Transformation Approaches

1. Obsolete methods: Traditional estimation methods (story points, planning poker) may die out as wasted effort in short iterations. The need for prescriptive frameworks may go down, which is reflected in the new kind of Scrum Master 2.0 role above.
2. Biodiversity: The era of company-wide standardization (aka. agile monocultures) may be kissed goodbye. Real-time KPIs keep leadership informed about what is going on. Remember that teams are complex entities and comparing/judging them based on KPIs is still a sin.
3. Learning-by-doing transformations: The clear winners of the “transformation war” are probably the light-weight transformation styles that allow organizations adopt AI-agility organically, step-by-step. Key elements: human beings smack in the middle, agile interactions and AI accessible data & tooling.

Assumptions: In the above list, I’m simplifying things by assuming AI to mean mere LLM (Large Language Model) without physical extensions. I’m assuming that AI’s development pace won’t accelerate in the next five years — if it does, then the changes to agility may be even more drastic.

Summary and Reflection

There is no rule book or manual for getting to 2030. We are learning as we go because the Agile-AI terrain is unclear. Organizations need to find the sweet-spot that unlocks efficiency, innovation and customer satisfaction without sacrificing the people-centric core tenet of agility.

At the moment many are singing AI-gospels, touting it as a “game changer”, maybe even “agile killer”. But is it, really? If we are blatantly honest, AI could end up as yet another shiny object that managers are running after. There has been so many: Virtual Reality (VR), Java Applets, Second Life, 3D TV, Flash, PalmPilot, Microsoft Bob, Windows Presentation Foundation (WPF), Silverlight, Microservices, Data lakes, Containers, Jira, Virtual Teams, Back-to-Office Mandates…

The absurdity of AI-hype is that it forgets human beings more than any new shiny object that came before. I’ve heard serious talk about having AI facilitate team events, even the holies of holy: retrospectives. 😳 To me, this sounds like a categorical fallacy and a gross misunderstanding of psychological safety.

What would AI do in a workshop if a hot conflict arises? I’ve seen people throw mousepads and mouses at each other! AI can’t intervene and it doesn’t have people skills. It can’t find calming or grounding words. It’s hard to express empathy/sympathy without a soul, you know. Luckily most agilists still recognize this, but for how long…?

This thought keeps me grounded. In order for AI to integrate with agile teams — as AI/DevOps kind of a thing — that team would need a proper CI/CD pipeline first. Although CI is a twenty-five and CD a fifteen year old ideas, not all today’s agile teams are there yet. So, we are far from even being able to integrate AI fully into our teams.

My bet until 2030 is that we will find some kind of middle-ground with AI that I call “Human-AI Bromance”. As the hype evaporates, AI will eventually be seen as a tool. Humanity regains its moral high-ground again. But before that happens, things can go south because risk happens fast. What a wild time to be alive!

Stay agile, people. Timo out.

Sources

• “Agile in the Age of AI” by Henrik Kniberg
• “AI, Methodology and Operating Model Evolution” by Jon Odo
• “AI in Agile Product Teams” by Stefan Wolpers
• Other snippets from here and there
• AI prompting
• Own thinking