Approved in Theory

Backend-developer-led teams optimise for what they can measure: tickets closed, features shipped, velocity maintained. It's rational behaviour inside a broken incentive structure.

Discovery work — prototyping, token architecture, design system foundations — doesn't fit in a sprint. There's no ticket for “create the future.” No velocity metric for “make the next five years faster.” So it gets deferred. Then deferred again. Then it becomes part of the mythical “next quarter” that never arrives.

I've watched Angular-to-React migrations get approved, then delayed, then deprioritised for three consecutive years. Not because React wasn't better. Because migrating doesn't close a feature ticket. The engineers who maintain Angular are productive. The engineers who would migrate it are a risk. The sprint board can't see what it's losing — only what it would cost.

This is the delivery trap. The team is working. They're shipping. Every metric looks fine. And the organisation is slowly calcifying around a codebase that the market left behind.

Here is the thing nobody says in the design systems community, because it makes us sound cynical: agentic design systems are politically dangerous.

A system that lets one person do the work of five isn't just efficient — it threatens headcount, budget, and organisational power. A CTO with 40 engineers and a £4M annual budget has power proportional to that team. A design system that compresses the need to 10 engineers and £1M isn't a technical upgrade. It's a political threat. The CTO will approve it in the boardroom and quietly ensure it never gets prioritised in the sprint.

This isn't cynicism. It's organisational behaviour. Quinn and Cameron's Competing Values Framework calls it Hierarchy culture — an organisation type that values control, stability, and the protection of existing structure above external adaptation. Hierarchy cultures don't reject innovation. They absorb it into committees and working groups until it becomes harmless.

The people who approve innovation are rarely the people who implement it. And the people who implement it are directly measured by the velocity that innovation would disrupt.

The innovation dies not from opposition. It dies from infinite, polite deferral.

Agentic design systems don't need a migration. They don't need the delivery team to stop what they're doing. They work alongside the existing codebase, from the inside out.

CLAUDE.md doesn't require a rewrite — it reads what's there and builds within it. A design token file doesn't replace the Angular components — it sits alongside them and makes the next thing built in React consistent from day one. You don't start with the migration. You start with one context file, one token, one component. The system grows from the inside while the delivery team keeps shipping.

The political answer is the same: stop asking for a migration. Start delivering outcomes that make the Angular codebase look obsolete by comparison. Build the NorthStar prototype in React that shows what the product could be. Let the output make the argument. The prototype argues better than any proposal deck ever could.

This is what the Stakeholder Simulator is built for. Not just to practice pitching — but to identify exactly which stakeholder is the actual political blocker, and what evidence they need to see before they'll unblock it. A CTO who fears headcount loss needs a different argument than a CTO who fears migration risk.

The delivery trap assumes that the team controls what gets built. Agentic design systems sidestep that assumption entirely. One architect with the right context files and a well-scoped agent can demonstrate in a week what a committee would defer for a year.