George's design maturity was growing, and we needed systems and rituals that could help the team operate efficiently at scale, and now, with AI in the picture too.
01
At George Labs I design for an 11M-user banking platform across 6 CEE markets. Alongside product work, I've led and hands-on shaped two initiatives that changed how our team operates: design ops and AI enablement.
I founded George's design ops function and ran it end-to-end, setting direction, facilitating the workshops, and shaping the artifacts myself. Four initiatives carried most of the weight.
Design quality and consistency varied across teams, with manual checks creating bottlenecks.
Initiated and shaped a Figma plugin to standardize workflows across Design and UXR. Currently exploring in-house AI Design QA agents to streamline validation further.
90% adoption rate on the Figma plugin, reducing manual effort and improving consistency at scale.
Design work was scattered across teams with no shared moment to surface it, learn from it, or align around it.
Organized Design Share, a recurring company-wide showcase, and facilitated hybrid workshops for 80+ designers across in-person and remote audiences.
64% consistent reach across the company, strengthening cross-team visibility and alignment.
Designers across markets needed shared standards and practical guidance to deliver consistent, high-quality work.
Co-shaped the UX Heuristics Framework and organized workshops on accessibility, Figma for devs, Blender, and AI tools.
85% of design OKR key results delivered for 12 consecutive quarters.
Knowledge was scattered, and new designers needed a structured way to ramp up.
Initiated and structured scalable Design Playbooks, Guidelines, and a structured Onboarding Experience covering Learning & Development, Figma Enterprise management, and AI design onboarding.
95% onboarding completion rate, 8.5/10 satisfaction. Materials people actually used and rated highly.
Design Ops outcomes
Across the programs, three signals mattered most. They spanned tooling, delivery, and people.
The Figma QA plugin became the default workflow across Design and UXR.
Of design key results delivered, sustained across 12 consecutive quarters.
New designers ramped up through structured playbooks, rated 8.5/10.
Helping the team build an AI practice, hands-on. Contributing to the program and shipping my own experiments inside it.
The spark
In December, I initiated and led our first AI design hackathon. 15 designers, one tool each, learnings shared as a team. The format worked, so we ran a second one a few months later.
The two of them sparked the practices that followed. Best practices, shared guidelines, and a regular cadence of AI experimentation took shape. Not as a finished system, but as ongoing processes the team keeps adding to.
Explorations
Currently exploring and experimenting with AI across different parts of the design workflow.
Copy changes meant updating each frame by hand, especially painful in user research where copy had to be adjusted for multiple markets.
Used Figma Make to update copy across flows with prompts.
Hours back per iteration, particularly in multi-market research where copy variants used to slow everything down.
Illustrations were produced manually. No AI in the workflow.
Explored Nano Banana, Config UI, and other tools to generate on-brand illustrations.
Illustrations now used in presentations and internal materials, including internal social media.
UX audits were manual and slow, with most of the time spent on preparation.
Using Claude to audit pages against heuristics in minutes and generate audit outputs.
A faster first pass that supports the review process. Still experimenting with prompts and outputs to find the right balance between AI surfacing and designer judgment.
Building the conference website usually meant 2-3 weeks of development plus months of ongoing support.
Designed the site in Figma. Used Claude to generate and deploy the HTML.
A working site shipped without spending development resources, freeing the engineering team for product work.
Reflection
AI enablement is still evolving. There's plenty of room to experiment, and the program is running, not finished.
What I'd do differently: schedule time daily to explore tools and see how they could improve the workflow. And manage expectations carefully, especially in a highly regulated environment, where adoption needs to move alongside compliance, not ahead of it.
Always open to a good conversation about financial health, AI x design, or behaviour change 💙