03 — AI · FinTech · Founding Design Partner

AI Invoice
Generator

Founding design partner on three AI features for Invoice Simple — diagnosing a low adoption problem rooted in discoverability and activation gaps, not satisfaction.

Company
Invoice Simple · EverCommerce
Role
Founding Design Partner (solo)
Features
Note Refiner · Text-to-Speech · AI Email
Year
2024

Design from scratch, no system to lean on

I was brought in as the founding design partner on Invoice Simple's internal AI initiative — a set of AI-powered features embedded in the invoicing product. No existing AI design patterns, no mature design system, no prior art to reference internally. Everything was built from first principles.

The three features in scope:

Low adoption — but not for the reason everyone assumed

Early data showed adoption of all three features was significantly below target. The instinct in the room was to attribute this to poor feature quality or user skepticism about AI. But the data told a different story.

"Users who found the features and tried them had strong satisfaction scores. The problem was almost no one was finding them."

This distinction mattered for what came next. A satisfaction problem requires redesigning the feature. A discoverability and activation problem requires redesigning the surfaces around it — entry points, empty states, progressive disclosure.

Post-launch unmoderated research on the Note Refiner

I ran unmoderated usability research specifically on the Note Refiner feature after launch. Three actionable findings surfaced:

[ Note Refiner — entry point before and after ]

Fixing the activation gap

The redesign work focused on three layers:

Working without a mature design system was a real constraint. I built reusable patterns and components for the AI features that I documented for the team — effectively creating a lightweight AI interaction library as part of the design work, not separately from it.

[ AI feature component library — entry states, loading, preview, error ]

Results

3
AI features shipped as founding design partner
Feature discovery rate post entry-point redesign
3
Research findings actioned in next sprint

The deeper outcome was a diagnostic framework the team could reuse — separating discoverability problems from activation problems from satisfaction problems — so future AI features could be evaluated with more precision.

← Previous Mobile Onboarding Redesign Next → Streamlabs Desktop