One language for 200+ applications
A design system for the plumbing of global finance, and the micro-tools each vertical actually needed.
Practice — Design systems · Enterprise UX · Financial infrastructure · Design governance
Client — The largest clearing and settlement corporation in the United States
Two hundred applications, one grammar.
An operations analyst at a clearing house moves between a dozen applications in a single shift. Same instrument, same trade, same afternoon — and a different vocabulary in every window. A confirmation is a confirmation here and something else two screens later. Nothing on any one screen is wrong. The cost lands in the gap between them, and it is a person under time pressure who pays it.
The client is the largest clearing and settlement corporation in the United States — an institution whose platforms underpin global markets, spanning multiple verticals. The capability was strong. What was missing was shared: no unified design language, no consistent internal tooling, no system that held across a diverse platform estate.
The brief came in as a queue. Optimise 200+ applications for operational efficiency. Application by application. Taken literally, that is a decade of work that ends where it started — two hundred well-designed applications that still don’t agree with each other.
I argued for the opposite order. Build one grammar first, then give each vertical only the micro-tools its own work needs. The system becomes the deliverable; the screens follow from it.
The rules we designed by
The system is the deliverable. Screens are its output.
One shared language everywhere; a custom tool only where the work genuinely differs.
In a regulated estate, consistency is a safety feature, not a style choice.
What we did
01 · Audited the estate as one system, not two hundred projects.
We catalogued what already existed across verticals and found the same problems solved several ways over. That inventory, not a redesign, was the first deliverable — and it was what made the case for a language.
02 · Built the grammar: foundations, components, patterns, governance.
Governance mattered more than the components. A component library without a rule for who may extend it becomes two hundred dialects again within a year.
03 · Built micro-tools per vertical, and refused the universal one.
A single system that fits nobody is worse than no system at all. Each vertical got the small tools its own work needed, built on the shared foundation rather than beside it.
04 · Defined where design sat — and what it was allowed to decide.
In a highly regulated ecosystem that positioning question is the real one. Design that cannot say no to a screen is decoration with a process attached.
How the work ran — DISCOVER · DEFINE · IDEATE · PROTOTYPE · DEPLOY & TEST. The double diamond, run linear or iterative depending on the size of the problem.
Project highlights
The challenge — Two hundred dialects
A strong platform estate with no shared design language, no consistent internal tooling and no system across verticals. At clearing-house scale, every inconsistency between applications is an operational risk carried by a person under time pressure.
The approach — Grammar before screens
One design system — foundations, components, patterns, governance — that every platform could build from, plus micro-tools tailored to individual verticals. And the positioning work that gave design standing inside a regulated environment.
The outcome — One language, held
Consistency across a diverse platform estate. Efficiency for the internal teams building on it. Clarity for the clients on the other side of it. A language the organisation could keep building on without me.
Gallery
The estate before intervention — every team solving the same problems differently.
The grammar — foundations, components, patterns and governance in one artefact.
Micro-tools, per vertical — built on the shared foundation, not beside it.
The same estate, unified — one language, held across every vertical.
Results — “A language that outlived the engagement.”
What changed for the user
The analyst moving between a dozen applications stopped relearning the interface at every jump. Same pattern, same meaning, same place on the screen, whichever vertical they landed in.
What it was worth
200+ applications brought under one design language.
Consistency across the estate. Efficiency for the teams building on it. Clarity for the clients on the other side of it. No measured figure exists here; these stay directional until one does, and the page says so rather than implying more.
The honest line
The durable test of a design system is whether it survives the person who built it. This one did. What I could not do from inside a vendor engagement was own the governance long enough to watch it hold — that stayed with the client.
200+ applications. One grammar. Held across every vertical.