Notes.
Notes. Some notes about me, some about my approach to design.
A bit about me
I’ve spent 12+ years embedded in corporate, government, and venture environments, working end-to-end across strategy, research and design lifecycles.
For the last ~2+ years I’ve been consulting through my own company with short- and long-term engagements, offering everything from fractional design leadership to being embedded in squads as an IC.
I’m often switching between design leadership and leading the work, where I’ll shape product and design direction while staying immersed in the problem space and seeing initiatives through to impact.
I’m drawn to design because of the universality of its application. Different sectors, different challenges, same potential; how do we connect pixels, people and profits.
Scaling with AI
It’s fascinating to see what continues to play out in this space. I love what I’m seeing, though I keep cautious optimism. Product design isn’t an obvious fit for AI (to me, anyway) given the role intangibles play when executing.
I’ve been experimenting (a lot) with different tools to better understand what actually unlocks value and where and what’s just a more messy way of getting to where you would have gotten anyway.
Unlocking velocity has been the more readily accessible value add, preserving context and keeping outputs consistent is the thing that continues to disrupt my workflows. I see this a lot in-house client-side too, in big and small organisation.
The balance between deterministic and probabilistic results is delicate. For me this has meant becoming very intentional with how and when I lean on AI tools. With the right guardrails, I’ve been able to rapidly explore ideas in high fidelity, shorten experiment loops, make faster decisions and condense the time I’d normally spend in synthesis and reporting.
I’ve learned how important it is to slow down the process at specific stage gates, otherwise these tools will swiftly truncate and optimise the work to a place you’re not very familiar with.
It’s important to remember the artefact/output itself isn’t the outcome, deep understanding, clarity and the ability to make better decisions is. That hasn’t changed. That means continuing to prioritise understanding and not substituting it for speed.
Current AI-design stack
Perplexity for market and competitor deep-research, exploring social sentiment on specific topics and asking why my hot-key isn’t changing my display brightness anymore.
Claude for sense-checking, ideation, synth, analysis and reporting. Also the occasional vent in voice mode.
Claude Code for prototyping and building. Fun-fact, I built this portfolio using Claude Code and live server by directing designs (as minute as border px width) in the HTML file. Did it take longer than if I had just used Figma? Yes. Did I learn new workflows that unlocked a lot of capability? Yes.
Figma Make for creating interactive components within frames without leaving the app. I love the way we can now just spin up complex components and quicken that whole experiment loop before committing.
Rovo when working client-side and needing to explore dense Confluence spaces (mostly client-side) in short periods of time. Side note, I’ve been really impressed with how efficient and consistent Rovo has been both as a collaborator and synthesiser.
Magic Patterns for converting a series of designs into a live demo I can put in front of clients or participants in user research.