Lead Product Designer · Financial Finesse · 2026

An AI authoring tool that scales financial content — without losing the voice that makes it trusted

A brand-trained AI workflow product replacing generic AI for the Financial Finesse content team. Designed and built end-to-end as a solo, AI-native designer — now powering financial wellness content for over 1M employees across Fortune 500 clients.

1M+
Employees reached
9 mo → weeks
Traditional vs. AI-native
Solo build
Designed + built end-to-end
Hero — full app screenshot (sidebar · editor · details panel)
My Role

End-to-end product ownership

Strategy

Identified the business constraint and defined the product opportunity — turning a content-team bottleneck into a workflow product, not a feature.

Design

UX, UI, workflow patterns, component library, and structured editors for all 9 content types — from research to high-fidelity.

AI Training

Built the brand voice, compliance, and content type schemas the model runs on. The intelligence layer that makes generic AI safe for FF.

Build

Coded the working product end-to-end using AI as my engineering partner. No spec doc handoffs. Design and build in the same loop.

Ship

Delivered a production-ready app to the team. Engineering came in only at the deploy stage to move it onto our internal infrastructure.

The Problem

A content team scaling for millions, working at the speed of one writer at a time

Content for over 1M employees across Fortune 500 clients moved at the speed of one writer in a Google Doc.
Generic AI couldn't fill the gap — wrong voice, banned phrases, drifting into individualized financial advice (a hard compliance line FF cannot cross).
9 content types · 5 audience segments · multiple regions — every dimension multiplied the load.
The Opportunity

What if AI didn't just generate content — what if it understood the voice, the rules, and the workflow of the team using it?

What if a writer could produce a compliance-safe, brand-aligned, audience-aware draft in minutes — and spend their time refining instead of writing from zero?

Generic AI generation Brand-trained, compliance-aware authoring
Drafts in scattered docs Library with status, owners, review flow
Manual voice + compliance review Ambient quality signals in the editor
One-size prompt Structured editors per content type
The Solution

From content factory to content system

I designed FF Content Writer as a workflow product, not a generator. The AI is trained on FF's voice pillars, compliance non-negotiables, banned phrases, audience tones, and the structure of all 9 content types. The interface wraps that intelligence in a real authoring experience — so the team doesn't fight the tool to make it useful.

01

A purpose-trained AI grounded in FF's voice pillars, compliance rules, and 9 content type schemas — every generation reflects the brand

02

Structured editors for articles, money tips, quizzes, coach insights, calculators, checklists, infographics, user stories, and videos

03

A real workflow — drafts, assignees, review notes, approve / send back, status tracking, and a "my queue" view

04

Ambient quality that flags missing CTAs, weak meta descriptions, banned phrases, and voice drift — with click-to-fix AI actions

The team stopped fighting the tool to keep the voice. The tool started defending the voice for them.

Experience · 01

AI that writes like a Financial Finesse coach — not like AI

Every generation runs through a system prompt built from FF's voice pillars, compliance rules, banned word lists, audience tones, and content type schemas. Drafts come back already aligned to FF's voice instead of needing rewrites.

Sidebar prompt → generated FF-voiced article in editor
Experience · 02

Nine content types, nine right-shaped tools

A money tips carousel doesn't need the same editor as a CFP coach insight or a calculator spec. Each content type gets its own structured editor with the right fields, the right AI prompts, and the right preview.

Multi-editor grid — article · quiz · checklist · money tips · coach insights
Experience · 03

Where drafts become a system

Every piece carries status, owner, audience, content type, and review notes. Filters cut by audience, status, and type. Multi-select bulk actions move five quizzes through review at once. "My queue" gives a writer only what's on them.

Library view — status chips · audience filter · bulk action bar · my queue toggle
Experience · 04

Quality signals you can't ignore — and can fix in one click

A right-panel checklist watches the draft in real time. It flags missing CTAs, weak meta descriptions, banned phrases, off-target reading level, and voice drift. Each signal is one-click fixable with a scoped AI action.

Live signals: CTA · meta · reading level · voice · banned phrases
Click-to-fix AI actions scoped to each issue
Compliance becomes ambient — not a final review step
Details panel — quality checklist with click-to-fix menu open
Experience · 05

Refine the part that needs it — not the whole thing

Highlight any sentence and pick Improve, Shorten, Simplify, Warmer, More actionable, or Ask. It replaces an old refine-scope model that nobody understood. Writers stay in flow.

Selection-based AI toolbar floating above highlighted text
Experience · 06

A pro tool that doesn't feel like a developer console

Autosave every 10 seconds. Drafts born with a name. Soft-delete with a 30-day Trash. Similar-content warnings as you type a topic. A coachmark the first time you hover the section AI button. Every detail tuned to the content team's actual day.

Details montage — autosave · similar-content warning · soft-delete · coachmark
Process

Designed and built solo. AI-native, end to end.

A traditional design-and-build cycle for a workflow product this deep takes a designer, a PM, and a multi-person engineering pod nine months. I designed and shipped it solo — using AI as my engineering partner — in weeks. Engineering came in only at the production handoff.

~9 months end to end Weeks to a working product
Designer + PM + multi-person dev pod Solo — designer as builder
Specs → handoff → revisions → build Design and build in one loop
Dev team needed from day one Dev needed only at production deploy

The product proves what AI unlocks for content teams. The way I built it proves the same thesis for design.

Impact

A tool the team ships with every day

1M+

Employees reached through generated content

9 mo → weeks

Traditional build cycle vs. AI-native solo build

Fortune 500

Trusted by Meta, NFL, JPMorgan, Federal Reserve

Meta NFL JPMorgan Chase CVS Health Federal Reserve Patagonia McKinsey Nestlé Comcast General Mills
What I'd do next

This work proved the model. Here's where it goes further.

01

Persistent backend + real auth

Move beyond browser storage so the team can collaborate from anywhere — with real version history, permissions, and a single source of truth.

02

Real-time co-editing & comments

Figma-for-content — multiple writers in one draft, threaded review notes, presence indicators, and a faster review loop.

03

Taxonomy-aware related content

Auto-detect duplicates and surface "refresh this existing piece instead?" before a writer starts a new one — the differentiating move beyond keyword search.

04

Mobile authoring

Capture coach insights, edits, and approvals from anywhere — meeting the team where the content actually happens.

Final Takeaway

FF Content Writer turned the content team's biggest constraint — voice consistency at scale — into a system. I built it the same way I designed it: AI-native, end to end, in weeks instead of months. Generic AI gives you words. A brand-trained workflow tool gives you a team that ships. A senior product designer who can build it gives you both — at speed.