Beyond individual productivity: rethinking AI strategy in product teams

What I learned about AI in design-dev collaboration through hands-on experiments and a 3-month thesis project, and why it led me to DesignOps.

“UX is dead.”

As a UX designer with a background in front-end development, I was struck by this phrase that kept popping up on LinkedIn. It also worried me a little for my future. But who exactly had killed UX?

The culprit seemed obvious: generative AI. Developers were embracing it faster than designers (or so I read). I realised that I used AI a lot when I was coding but barely touched it in my design work. Meanwhile, my developer colleague, an AI power user, was releasing features faster than ever.

This observation led me to a question that became the starting point of my thesis at Les Gobelins Paris: How could I, as the only designer on my team, keep pace and stay relevant in this changing environment?

The research that changed my mind

I spent three months, from June to August 2025, actively investigating this question through my thesis. Many resources have shaped my understanding of AI and collaboration, from books such as The Design Conductors to articles like Dev-Design Collaboration Framework or How AI Tools Are Bridging the Gap Between Product Thinking and Frontend Code.

Building on these foundations, my approach was deliberately comprehensive. For internal research at my company, I conducted AI experiments, tried various tools, and led three field tests at three crucial collaboration stages: scoping, prototyping and handoff. For external validation, I ran an online survey to which 28 designers answered, and conducted 6 interviews with designers and developers.

But the results surprised me. I expected to confirm my fears, developers racing ahead while designers fell behind. However, this observation wasn’t validated. That’s when I thought: Do designers really need to be faster and more productive?

I understood that before finding ways to adapt, I needed to understand what AI really had changed between designers and developers. I refined my angle to a more relevant one: How is generative AI reshaping collaboration between designers and developers, from the scoping phase to the handoff?

This reflection ultimately led me to view AI less as an isolated tool and more as an organizational topic (DesignOps).

What I discovered

Allow me to introduce the three members of the ‘product trio’ that will assist me in presenting my results:

• Marie, a passionate designer who is curious about AI
• Tony, an astute developer who’s also an absolute fan of AI
• John, Product Manager, who loves collaborating with his team

1. AI makes individuals faster, not teams better

Marie represents 77% of designers using AI in my survey who reported speed gains since adopting it. She uses ChatGPT for interview guides and Figma Make for rapid prototyping. In one test, I reduced my prototyping time from two days to two hours using AI.

What I found interesting: Marie’s relationship with Tony didn’t change at all. Both my data and interviews confirmed this pattern. AI frees up individual time, but teams rarely reinvest those savings into better collaboration. It’s rather used to focus on tasks that are seen as more valuable, like understanding users’ needs or developing strategic thinking.

2. AI augments rather than replaces (for now)

When Marie tried AI prototyping, she hit walls quickly. The tool sometimes ignored parts of her design system, created visual inconsistencies, and missed her design intentions. Tony experienced similar frustrations with AI-generated code that was more difficult to understand and failed to meet the company’s or his own quality standards.

In my handoff experiment, I tested how a developer approached a traditional Figma file versus one created with Figma MCP + Cursor. The developer I worked with found the traditional approach clearer and easier to interpret.

What became clear, though, was AI’s potential as a thinking partner. For the scoping phase, I conducted a “synthetic developer” test (an idea based on the synthetic user experimentations), using ChatGPT to simulate developer feedback during scoping sessions. While it couldn’t match my real colleagues’ depth, it helped me prepare better questions and spot blind spots I’d missed.

John, our PM, found similar value in AI prototyping. Even imperfect prototypes helped our ‘product trio’ to iterate and align more quickly, as well as develop a shared language around complex ideas.

3. Adoption is prevalent, but lacks strategy

Marie isn’t unique. Only 2 of 28 survey respondents said they don’t use AI at all. But adoption is happening in an exploratory phase mostly without global strategy:

• Some designers feel pressured to use tools that clash with their values

“I’d say it’s more pressure. I’ve been a bit resistant to AI because I don’t find that the efficiency outweighs the environmental cost at the end of the day. But I’m slowly changing my mind and I’m trying to use it for very specific things.”
— 7th participant to user interviews (Product Designer)

• Others received conflicting instructions on how to use AI, with designers being banned from using it and developers being encouraged to use it, and vice versa
• Others describe frequent discussions about AI and tools, but no capitalisation on what was learnt through individual experimentation

Companies are aware this is a key issue but implement solutions piecemeal. The result is fragmented practices, missed learning opportunities, and unequal access across teams.

A different perspective on change

What struck me most during this research is that AI isn’t fundamentally changing collaboration: it makes existing collaborative fundamentals even more crucial.

Good collaborators can leverage AI more effectively. Teams with poor communication don’t magically improve just because they can prototype faster. Establishing frameworks that make designer-developer partnerships work becomes more important, not obsolete. The framework designed by Alicia Calderón González (communication, collaboration, alignment), Lead Design Strategist at Miyagami, which I discovered later in my research, seems even more relevant today.

Moving Beyond Individual Adoption

Based on my findings, I developed 11 recommendations using Nielsen Norman Group’s DesignOps resources. You’ll find under some examples. The key insight: we need to evolve from individual AI adoption to team-wide transformation.

Governance:

• Audit existing employee practices: Individual practices have already evolved to include AI, but in silos. Understanding current usage patterns helps identify where AI has already penetrated processes.

Process:

• Create an AI usage charter: A shared framework helps align practices between designers, developers, and across the entire company.
• Establish guided experimentation: Create sandbox environments for testing AI in collaboration without impacting critical deliverables. Tools that bridge both professions offer particularly interesting opportunities for joint designer-developer experiments.

Operational Support:

• Invest in shared training: Provide designers and developers with common foundations in prompting, technical basics, and chosen tools to create shared language and collective skill development.
• Enable effective documentation: Document what works and what doesn’t to define and evolve AI workflows based on experimental results.

Bear in mind that AI is still in the exploratory stage and prone to instability. My goal here is not to offer an absolute framework of how things should be, but rather to provide actionable recommendations that will help to move beyond individual or silo adoption. The aim is to use AI to facilitate collaboration.

Looking forward

This research had its limitations (small survey sample, context-specific tests, timing constraints), but comparing my findings with larger studies like Figma’s State of the Designer 2025 revealed consistent patterns across different contexts.

What I’ve observed is that teams are currently in an exploratory phase with AI, often lacking strategic coordination. The teams that will benefit most aren’t necessarily those who adopt AI fastest, but those who think most carefully about how it fits into their collaborative practices.

AI won’t replace design or development roles, but it is creating new opportunities for teams willing to experiment thoughtfully with how they work together.

What’s your experience with AI in product teams? Have you seen it improve or complicate collaboration?

Many thanks to all who contributed their time and insights to this research. Interested in exploring the recommendations in detail? Feel free to reach out to discuss!

Beyond individual productivity: rethinking AI strategy in product teams was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.

This post first appeared on Read More