Design systems don't die from bad design
They die in meetings. A CEO who doesn't see the revenue impact. A CTO who doesn't trust the architecture. A CFO who sees cost without return. A product manager who can't fit it in the roadmap.
The design systems architect who can only talk to designers has already lost. The one who can walk into a C-suite meeting and translate “multi-brand token architecture” into “we launch new brands in days instead of months” is the one who gets funded.
Alan Cooper called this role the Design Communicator in About Face. Someone who translates between the design vision and the business reality. The problem is: most design teams don't have one. And the architect usually isn't trained for it.
Simulate the room before you enter it
I built a Stakeholder Simulator agent that inhabits different executive perspectives. I pitch my idea, and the agent responds asthe CEO, the CTO, the CFO — pushing back, asking hard questions, finding the gaps in my argument.
CEO
“Will this make us win?”
Revenue, market position, competitive advantage, board narrative
CTO
“Will this scale? What breaks?”
Architecture, scalability, tech debt, team capacity
CFO
“What does this cost?”
ROI, headcount, budget cycles, payback period
COO
“How does this affect operations?”
Timelines, process impact, change management, disruption risk
Engineering Lead
“Will my team adopt this?”
Developer experience, learning curve, migration cost, velocity
Product Manager
“Where does this fit in the backlog?”
Roadmap alignment, sprint planning, stakeholder management
Simulated pitch: agentic design systems
Here's a simulated conversation where I pitch agentic design systems to a CEO and CTO. The agent plays both executives. I respond in real-time.
So you want to invest in what you're calling an 'agentic design system.' Give me the thirty-second version.
One system that serves every brand we own. AI agents consume the tokens and components directly. We cut design-to-delivery time by 80% and reduce the team needed for each new brand from five people to one.
Hold on. 'AI agents consume the tokens' -- what does that actually mean technically? Are we talking about an MCP server? Custom tooling? Or just a markdown file and a prayer?
CLAUDE.md context files, design tokens as structured JSON, Code Connect mapping Figma to codebase, and MCP endpoints the agents can query. The design system becomes API infrastructure, not a Figma library.
OK, that's more concrete than I expected. What happens when the agent hallucinates a component that doesn't exist?
The token contract constrains it. Typed variants, structured props. The agent can't invent a button variant that isn't in the system. The design system IS the guardrail.
I like the 80% number. But I've heard 'design system will save us time' before. Last time we spent eighteen months building one and the developers ignored it.
That's because it was a Figma library, not code. Developers don't consume Figma. This system starts in code and flows to Figma, not the other way around. The developers are the first users, not the last.
And the AI part -- is this real or is this a conference slide?
My portfolio site is the proof. Every page was built by AI agents consuming the design system. The site demonstrates the methodology it describes. I can show you the agent definitions, the token architecture, and the deployed result.
The N in I<N>C is for every audience
The Narrate layer isn't just machine-readable context for AI agents. It's the ability to encode and communicate vision for anyaudience — CEO, CIO, CTO, PO, SME. Same truth, different language.
This maps to the MIND framework's Narrative Reasoning— thinking in stories, simulating conversations, encoding complex ideas through narrative. The Stakeholder Simulator is that cognitive function made operational.
Most design systems architects can explain their system to other architects. The ones who change organisations can explain it to anyone in the room. The simulator lets me practice that translation before the stakes are real.
The best ideas don't win because they're best. They win because someone could explain them to the person holding the budget.
Alan Cooper meets agentic AI
Cooper identified the Design Communicator as a missing role in software teams — someone who translates between the design vision and the business. Most companies still don't have one.
AI changes this. You don't need to hire a Design Communicator. You need to build one. An agent that knows your design system, knows your positioning, and can simulate how a CFO, a CTO, or an engineering lead would respond to your proposal.
This is also how I lead my AI team. I have four lead agents — UI Designer, UX Designer, FED Developer, UX Researcher. The Stakeholder Simulator is the fifth. It doesn't make things. It makes sure the things we make survive contact with the boardroom.
“Design systems don't need better components. They need better communicators. The agent simulates the audience so the architect can practice the translation.The Stakeholder Simulator thesis
Want to pressure-test your design system pitch?
Whether you're pitching to a CEO, defending to a CTO, or justifying to a CFO — I've built the practice room. Let's talk about how narrative reasoning changes how design systems get funded.