Generations of wealth, inheriting a legacy UI

The brief

Archway manages general ledger infrastructure for ultra-high-net-worth family offices. Their client-facing investment portal had been built iteratively over several years. The ask was aesthetic modernization. The real problem was something else entirely.

What we found

Before touching a single component, I pulled engagement analytics on the existing portal.

The portal had been designed for both the wealthholder and the family office operator without ever committing to either. The information architecture reflected internal workflows, not client goals. It was too slow and basic for ops, and the wealthholder used it four times a year to get documents.

The product wasn't broken. It was aimed at no one in particular.

The reframe

I brought the analytics to stakeholders with a single provocation: who do we actually want using this portal, and what do they need to do when they get here?

The answer was the wealthholder. Not the operator managing the ledger. The person whose wealth it is, reviewing performance, understanding allocation, tracking distributions. The redesign brief changed. The goal was no longer just aesthetic modernization. It was to build a portal that could earn the wealthholder's trust and become a genuine part of how they stay connected to their wealth.

Process

Audit and alignment. Mapped the existing portal against user type. Pulled engagement data to quantify what was already visible qualitatively. Facilitated a reframe with product and engineering leads, defining the wealthholder as the primary user and mapping the workflows that mattered to them: portfolio overview, document access, bill approvals.

From a single screen to a working prototype. Rather than building out a full Figma file to validate direction, I designed a single key screen — the primary dashboard view — and fed it directly into Claude with context about the user, the data model, and the interaction goals. Claude helped surface edge cases, stress-test the information hierarchy against real data patterns, and draft the interaction logic before a line of production code was written. The output of that session — user definition, data model, hierarchy decisions, interaction logic — effectively was the PRD. We never wrote one separately.

From there I used Figma Make to generate a working browser prototype from that single screen, letting stakeholders and engineering pressure-test the real experience rather than click through static frames. This compressed what would have been multiple rounds of lo-fi iteration into a single validated artifact.

Figma production and handoff. With direction validated, I translated the prototype into a full-state Figma file covering the dashboard, data states, empty states, and responsive behavior. Established the component architecture as a foundation that can extend into the operator experience without a persona conflict or a rebuild. New for me was also putting together a .md file to keep consistency in future AI prototyping sessions.

Artifacts

• Legacy portal audit
• Single-screen concept to Claude-assisted prototype
• Figma Make browser prototype
• Production Figma file — full state coverage, design system foundation

Full design files available on request.

Outcomes

Stakeholder alignment on wealthholder-first product direction — a strategic shift from the original brief.

New dashboard in active development with engineering, built on the redesigned foundation.

Component architecture established for extension into operator workflows.

What's next

Phase 1 is in development. The next phase focuses on high-frequency workflow loops and begins collapsing the gap between what the portal offers and what's currently being done in Excel.