Making AI work for agile teams: research roadmap from XP2025

The conference panel at XP 2025 was not a typical one; over 30 contributors, including agile pioneers and AI specialists, academic and industry leaders gathered from all over the world. Their aim was to move beyond the buzzwords and get honest about the challenges agile teams are facing today. My colleague Munish Malik and I had the privilege of joining this forward-thinking event, a highlight of which was a full-day workshop titled ‘AI and Agile: From Frustration to Success’, where we explored how generative AI is influencing software delivery, and what’s standing in its way. Together, we built a research roadmap aimed at enabling responsible, effective AI integration in software delivery.

Facing the friction: what we heard

The workshop combined keynotes, experience reports, and interactive breakout sessions designed to tap into real-world practice. Participants shared how their teams were experimenting with AI—what was promising, what flopped, and what still feels unresolved.
The discussions were lively and grounded. We collected over 120 pain points, clustering them into themes ranging from tool fatigue to a lack of AI literacy. Some teams were overwhelmed by too many disconnected tools. Others struggled with ethical concerns and gaps in governance. And there were plenty of stories about missed opportunities due to a lack of prompting skills or underwhelming AI creativity.
These weren’t just complaints. They were insights. By the end of the day, we’d turned this raw data into a focused research agenda, designed to balance immediate needs with long-term aspirations.

Where AI in agile stands today

Here are three things that stood out for us in the research:

1. The potential is real—but the ecosystem is fragmented.
   There’s a flood of AI tools on the market, but few integrate smoothly. For agile teams, that means wasted effort stitching together systems that don’t talk to each other.
2. Skills matter just as much as the tech.
   Even the most powerful AI tools won’t deliver value if teams don’t know how to use them effectively. Prompting, understanding data flows, and interpreting outputs are now critical skills.
3. Governance is urgent.
   Concerns around data privacy, intellectual property, and regulation are already slowing AI adoption. Without addressing these head-on, trust in AI will remain brittle at best.

A roadmap for smarter AI adoption

The main outcome of the workshop was a five-part research roadmap. It’s not a final answer—but it’s a practical place to start.

1. Tooling ecosystem and integration
   Tackling the overwhelming fragmentation of AI tools by developing clear selection frameworks and unified interfaces. Our goal is reducing cognitive overhead while creating seamless integration solutions that actually boost team productivity.
2. Human factors and AI literacy
   Building the skills, understanding, and mindset needed to work with AI collaboratively. Developing essential prompting capabilities, effective onboarding strategies, and role-specific approaches that recognise each team member’s unique needs in collaborative human-AI relationships.
3. Governance and compliance
   Creating experimentation environments where we can explore AI capabilities responsibly while maintaining compliance standards.
4. Value realisation
   Moving past the initial AI hype through rigorous evaluation methods and continuous feedback systems. Defining success, tracking impact, and ensuring AI brings measurable benefits.
5. Creativity and multimodality
   Investigating how multimodal AI can enhance our design thinking, ideation processes, and innovative problem-solving within agile workflows. The focus is integrating these capabilities while preserving the human elements that drive breakthrough innovations.

Download the full XP2025 workshop report to explore the findings and join the conversation. The full research paper linked here expands on these five themes, and provides a structured framework for addressing the complex challenges of AI integration in agile software development.

This roadmap is an open invitation for collaboration. It’s a starting point for researchers, practitioners, and organisations—Equal Experts included—to define and build responsible, human-centred AI practices.

Let’s shape what’s next

Generative AI is here to stay. But how it supports agile software delivery is still being shaped. What we saw at XP2025 made it clear: the future of AI isn’t just about smarter tools. It’s about smarter teams, better questions, and systems we can trust.
If you’re exploring how AI fits into your agile practice, we’d love to hear from you.

About the authors

Munish Malik is a Product Principal at Equal Experts with nearly two decades of experience helping teams turn ideas into products people value. He enjoys working in discovery-led ways, blending strategy, design thinking, and engineering to solve the right problems. Munish has also led AI-accelerated initiatives that use LLMs to speed up legacy system discoveries and deliver business impact.

Mohit Chandna is a Product Consultant at Equal Experts with over 14 years of experience solving customer problems across diverse domains, including payments, healthcare, banking, and security. A trusted advisor and prominent community voice, his curiosity now focuses on the evolving world of Generative AI, where he is actively discovering the details. He organizes ProductTank meetups in Delhi NCR and has spoken at major conferences like Agile on the Beach UK, Agile Manchester, Agility Today India, the Atlassian Community Event, and the XP 2025 conference.

Disclaimer

The views shared in this blog reflect the personal experiences of Mohit Chandna and Munish Malik at the XP2025 workshop. They do not represent Equal Experts’ official practices or methodologies.