Integrating AI augmented workflows for design & business.
The design process has always been about making better decisions faster. Over the rejent past, I have integrated AI tools into my workflow — not to replace design thinking, but to amplify it. AI helps to move from research to validated design directions more quickly, explore more variations than would be practical manually, and create richer conversations with product managers and stakeholders.
UX Research — Synthesis and Question Design
One of the most time-consuming parts of UX research has always been synthesis — taking raw interview data, survey responses, and notes, and identifying the patterns that matter. AI has fundamentally changed how quickly and thoroughly I can do this.
I use Claude to help design user interview scripts and survey questions, working on both a qualitative and quantitative basis. Rather than starting from a blank page, I collaborate with Claude to generate question sets tailored to specific research goals, then refine them based on my domain knowledge and the nuances of the project. This cuts what was previously a full day of preparation down to one or two focused hours.
After conducting interviews or collecting survey data, I use Claude to process and synthesise the findings. I feed in transcripts and notes, and Claude helps identify emerging themes, patterns, and contradictions across participants. This does not replace my own analysis — I still make the interpretive decisions about what matters and why — but it dramatically accelerates the initial pass through large volumes of qualitative data, allowing me to spend more time on the insights that will actually shape the design direction.
Design in Figma — AI-Powered Prototyping and Variation
One of the most impactful additions to my workflow has been connecting Claude Code to Figma through MCP (Model Context Protocol). This allows me to work with existing Figma designs programmatically — I feed Claude Code the URL of a Figma file, and from there I can generate coded prototypes, create design variations, and make iterative changes far more rapidly than working manually within Figma alone.
The primary use case is rapid variation generation. Where previously I might produce one or two variations of a design for stakeholder review, I can now generate multiple variations in the same timeframe.
This has a significant impact on the quality of design conversations with product managers. More options on the table means richer discussion, more informed feedback, and better alignment on direction — even when not every variation is an improvement, the breadth of exploration satisfies stakeholders and provokes the kind of dialogue that leads to stronger outcomes.
Quick Prototyping for Stakeholder Feedback
Beyond static design variations, I use this workflow to create quick coded prototypes that I can put in front of product managers and other stakeholders for early feedback. Rather than describing how an interaction might work or walking through a series of static screens, I can showcase a working prototype that communicates the intent far more clearly. This accelerates the feedback loop and reduces the risk of misalignment between what was designed and what stakeholders expected.
Design Documentation
I also use Claude Code’s Figma integration to generate and maintain design documentation. Keeping documentation accurate and up to date has always been one of the less glamorous but essential parts of design system work. Automating portions of this process means documentation stays current without consuming the hours it previously required.
Code and MVP Development From Design to Working Code
Using Claude Code alongside Cursor, I can take designs and translate them into working code far more quickly than a traditional handoff process allows. This is particularly valuable for building MVPs and proof-of-concept prototypes where speed matters more than production-grade polish. Being able to move fluidly between design and code means I can validate ideas in a working environment earlier in the process, catching feasibility issues and interaction problems that are invisible in static mockups.
What AI Does Not Replace
I want to be clear about what AI does and does not do in my workflow. AI does not make design decisions. It does not understand the business context, the political dynamics of a stakeholder group, or the subtle frustrations of a user who has been working around a broken workflow for years. It does not replace the judgement that comes from spending two or three months learning a domain before proposing changes, as I did with Crusher Mapper at Metso.
What AI does is remove friction from the parts of the process that benefit from speed generating options, synthesising data, producing documentation, and building prototypes. This gives me more time and energy for the parts that require human judgement: understanding the real problem, making design decisions with incomplete information, and bringing people together around a shared direction.