BLOG

Thinking Out Loud

My thoughts on AI product strategy, design operations, and what actually changes about building software once agents are in the loop.

FEATUREDDesign Leadership

Why Design Systems Become the Control Layer for AI-Native Teams

I wrote this article while preparing my application for Puffy’s Head of Design, AI-First role. It sets out how I would help the team ship better designs, faster—through coded prototypes, shared design systems and a culture of learning.

VIEW
Diagram showing Figma and the design-system MCP connecting trusted system context into AI workflows.
The filled gap: design context from Figma and system guidance from the MCP reach the AI workflow together.
02 ·

AI enablement for designers must begin with the working environment

AI enablement for designers cannot stop at training. When tools, agents, and workflows change every few months, designers need a persistent working environment: a playground they can return to, adapt, and use to build real fluency.

AI Enablement
04 ·

Communities are adoption infrastructure, not an event calendar

Internal design communities create value when events connect people to follow-up paths, reusable resources and adoption signals.

DesignOps
05 ·

Design files now have two audiences: people and machines

AI tools now read Figma files as data. A structured starter file helps people and machines find the current, reusable, ready-to-build design.

Design Systems
03 ·

Enterprise design systems fail at adoption, not awareness

Enterprise design systems rarely fail because teams have not heard of them. They fail when teams cannot adopt them in real delivery work: access, component choice, design-code alignment, edge cases, and feedback loops.

Design Systems
06 ·

Global DesignOps only works when it becomes local

BMW Group had mature global DesignOps services, but growing teams at BMW TechWorks India still needed a local way to find and use them. This post explains how the UX Live Center turned a central service catalogue into a visible front door, a weekly support rhythm, and a feedback loop back to global teams.

DesignOps
07 ·

UX maturity should guide investment, not grade teams

UX maturity work is most useful when it decides where support goes. At BMW TechWorks India, the index became a way to route training, standards, measurement, sponsorship, and hands-on UX help, not a scorecard for ranking teams.

DesignOps
01 ·

What changes when design-system knowledge enters the AI workflow

An AI assistant can build a settings page in seconds. If your design-system knowledge is not available in that moment, it will quietly invent components, spacing, and patterns that look plausible but drift from the system.

Design Systems