AI-native product design leader

Velocity.
Systems.
Impact

Hello, I'm Yvonne Doll. As a product design leader, I lead by building. I set vision, align EPD, staying hands-on to turn complex AI product ideas into working prototypes and scalable systems. With deep experience in AI, B2B, CRM, SaaS, and remote leadership, I drive results. In my free time, I make music and am an abstract oil painter.

hello@ydoll.me

See my work

Leading by building

Read the full case study →

01GUIDE

Design rules + reusable skillsGUIDE 01

Codify Design Judgment

Answers: What should guide the work?

Make design intent, standards and repeatable methods available where the work happens. Give AI the context to help, and people a shared basis for review.

## What the guidance contains

R-07Destructive actions state the blast radius before confirming.

R-19Error copy states what happened, then what to do. Never “Something went wrong.”

R-24Tables over 50 rows ship with filter and column sort.

R-31AI suggestions stay visually distinct until the user accepts them.

## Output

A reusable reference for creating and reviewing work against design intent.

Instructions guide the work. They do not prove the result meets the standard.

02CONSTRAIN

Tokens, components + guardrailsCONSTRAIN 02

Make the Right Thing Easier

Answers: Which decisions should the system carry?

Build repeatable decisions into tokens, components and configuration. Make approved patterns easy to reuse, and define the limits of what can change.

## What belongs in the system

Tokens — shared values instead of disconnected styling decisions. · Components — reusable behavior and approved variants. · Configuration — flexibility with explicit boundaries. · Guardrails — warnings or enforced limits, depending on the risk.

## Output

A choice like primary, instead of raw CSS.

GOOD<Button variant="primary" size="medium" />

NOTbackground: #4263EB; padding: 12px 18px; border-radius: 6px

03VERIFY

Checks + AI-assisted reviewVERIFY 03

Check the Work, Then Investigate

Answers: Where does the output diverge from the intent?

Use automated checks for measurable requirements and AI-assisted review to surface potential design and copy issues. Validate findings against the code and the running experience.

## What a useful finding includes

Fail: R-07Destructive action lacks blast radius

Found“Are you sure?”

Fix“Update this tool? 5 procedures will be affected.”

Fail: R-19Generic error copy ×3

Found“Something went wrong”

Fix“Couldn’t save the tool. Check the required fields and try again.”

Pass: R-02Contrast ≥ 4.5:1 — 48 / 48 text nodes

Pass: R-31AI suggestion styled distinct — 12 / 12

## Output

A reviewable list of findings, with confirmed issues separated from suggestions that need investigation. A passing check covers what it tested. It is not a guarantee of overall UX quality.

04GOVERN

Decision rights + triageGOVERN 04

Keep Judgment Human

Answers: Who decides what happens next?

Set clear boundaries for automation and ownership for decisions. Weigh user impact, confidence and reversibility before choosing what to fix, escalate or defer.

## Decision boundaries

Automate — repeatable actions within explicit permissions and tested limits.

Escalate — uncertain or high-impact findings for human review.

Defer deliberately — record an owner, rationale and follow-up for accepted work.

## Learning loop

Finding → Validate → Fix, escalate or track → Review the outcome → Update guidance

Organizational Impact

WordPress/Jetpack | Kustomer

As Head of Design at Kustomer. com, I led the design org through a transformation to native AI cross-functional workflows, shaving concept-to-validation time from days to hours through streamlined collaboration,

At WordPress (Automattic), I spearheaded five AI launches for 27M+ users across 14 products, achieving a 58% productivity boost. I also designed and implemented a lean research program, which was widely adopted by the company.

SEE Organizational Impact →

Craft

VARIOUS

Designed the complicated workflows behind AI agent setup, guardrails, procedures, monitoring, and human review, turning an open-ended configuration problem into a guided, trustworthy experience for both end users and administrators tasked with setting up the AI agents. With experience in SaaS, healthcare, publishing and website security, I know how to design for trust.

Explore the craft →

Range w/a point of view

I’m a hands-on product design leader who brings executive judgment, creativity, and systems thinking to complex work. Find out more about me in this short intro video.

Thought Currents

All blog posts →

Zero UI

A thought excercise UX teams

Rabbit/Sloth hybrid, a Slobbit if you will

AI: The death of the handoff

Breaking the grid with Kim.

The Museum of Abandoned AI Features

Welcome to the Museum of Abandoned AI Features, where every exhibit began with the sentence, ‘Could we put a copilot in it?’

Work

Impact

Thought Currents

About

Resume ↗

Contact me

Velocity.
Systems.
Impact.

Craft.

Velocity.

Vision.

CHICAGO, IL · OPEN TO TRAVEL OR RELOCATE

FIG. 01 · THE ELEPHANT IN THE VIBE CODE

HELLO@YDOLL.ME

CHICAGO, IL · OPEN TO TRAVEL OR RELOCATE

FIG. 01 · THE ELEPHANT IN THE VIBE CODE

HELLO@YDOLL.ME

AI-Native: Leading
by building

AI-native: Leading by building

I moved Kustomer’s design team from Figma-first handoffs to an AI-native practice, working alongside product and engineering to shape and build the experience together.

I stay close to the work: untangling UX problems, building prototypes, and working with the team. That hands-on practice informs how I coach designers, raise the quality bar, and help the team move faster.

I stay close to the work: untangling UX problems, building prototypes, and refining the details. That hands-on practice informs how I coach designers, raise the quality bar, and help the team move faster.

Read the full case study →

~hrs

Idea to Prototype

↑40%

design capcity (by points)

01GUIDE

Design rules + reusable skillsGUIDE 01

Codify Design Judgment

Answers: What should guide the work?

Make design intent, standards and repeatable methods available where the work happens. Give AI the context to help, and people a shared basis for review.

## What the guidance contains

R-07Destructive actions state the blast radius before confirming.

R-19Error copy states what happened, then what to do. Never “Something went wrong.”

R-24Tables over 50 rows ship with filter and column sort.

R-31AI suggestions stay visually distinct until the user accepts them.

## Output

A reusable reference for creating and reviewing work against design intent.

Instructions guide the work. They do not prove the result meets the standard.

02CONSTRAIN

Tokens, components + guardrailsCONSTRAIN 02

Make the Right Thing Easier

Answers: Which decisions should the system carry?

Build repeatable decisions into tokens, components and configuration. Make approved patterns easy to reuse, and define the limits of what can change.

## What belongs in the system

## Output

A choice like primary, instead of raw CSS.

GOOD<Button variant="primary" size="medium" />

NOTbackground: #4263EB; padding: 12px 18px; border-radius: 6px

03VERIFY

Checks + AI-assisted reviewVERIFY 03

Check the Work, Then Investigate

Answers: Where does the output diverge from the intent?

Use automated checks for measurable requirements and AI-assisted review to surface potential design and copy issues. Validate findings against the code and the running experience.

## What a useful finding includes

Fail: R-07Destructive action lacks blast radius

Found“Are you sure?”

Fix“Update this tool? 5 procedures will be affected.”

Fail: R-19Generic error copy ×3

Found“Something went wrong”

Fix“Couldn’t save the tool. Check the required fields and try again.”

Pass: R-02Contrast ≥ 4.5:1 — 48 / 48 text nodes

Pass: R-31AI suggestion styled distinct — 12 / 12

## Output

A reviewable list of findings, with confirmed issues separated from suggestions that need investigation. A passing check covers what it tested. It is not a guarantee of overall UX quality.

04GOVERN

Decision rights + triageGOVERN 04

Keep Judgment Human

Answers: Who decides what happens next?

Set clear boundaries for automation and ownership for decisions. Weigh user impact, confidence and reversibility before choosing what to fix, escalate or defer.

## Decision boundaries

Automate — repeatable actions within explicit permissions and tested limits.

Escalate — uncertain or high-impact findings for human review.

Defer deliberately — record an owner, rationale and follow-up for accepted work.

## Learning loop

Finding → Validate → Fix, escalate or track → Review the outcome → Update guidance

01GUIDE

Design rules + reusable skillsGUIDE 01

Codify Design Judgment

Answers: What should guide the work?

Make design intent, standards and repeatable methods available where the work happens. Give AI the context to help, and people a shared basis for review.

## What the guidance contains

R-07Destructive actions state the blast radius before confirming.

R-19Error copy states what happened, then what to do. Never “Something went wrong.”

R-24Tables over 50 rows ship with filter and column sort.

R-31AI suggestions stay visually distinct until the user accepts them.

## Output

A reusable reference for creating and reviewing work against design intent.

Instructions guide the work. They do not prove the result meets the standard.

02CONSTRAIN

Tokens, components + guardrailsCONSTRAIN 02

Make the Right Thing Easier

Answers: Which decisions should the system carry?

Build repeatable decisions into tokens, components and configuration. Make approved patterns easy to reuse, and define the limits of what can change.

## What belongs in the system

## Output

A choice like primary, instead of raw CSS.

GOOD<Button variant="primary" size="medium" />

NOTbackground: #4263EB; padding: 12px 18px; border-radius: 6px

03VERIFY

Checks + AI-assisted reviewVERIFY 03

Check the Work, Then Investigate

Answers: Where does the output diverge from the intent?

Use automated checks for measurable requirements and AI-assisted review to surface potential design and copy issues. Validate findings against the code and the running experience.

## What a useful finding includes

Fail: R-07Destructive action lacks blast radius

Found“Are you sure?”

Fix“Update this tool? 5 procedures will be affected.”

Fail: R-19Generic error copy ×3

Found“Something went wrong”

Fix“Couldn’t save the tool. Check the required fields and try again.”

Pass: R-02Contrast ≥ 4.5:1 — 48 / 48 text nodes

Pass: R-31AI suggestion styled distinct — 12 / 12

## Output

A reviewable list of findings, with confirmed issues separated from suggestions that need investigation. A passing check covers what it tested. It is not a guarantee of overall UX quality.

04GOVERN

Decision rights + triageGOVERN 04

Keep Judgment Human

Answers: Who decides what happens next?

Set clear boundaries for automation and ownership for decisions. Weigh user impact, confidence and reversibility before choosing what to fix, escalate or defer.

## Decision boundaries

Automate — repeatable actions within explicit permissions and tested limits.

Escalate — uncertain or high-impact findings for human review.

Defer deliberately — record an owner, rationale and follow-up for accepted work.

## Learning loop

Finding → Validate → Fix, escalate or track → Review the outcome → Update guidance

Shaping product direction

WordPress/Jetpack | Kustomer

I lead design teams, shape product strategy, and stay hands-on with the work. Partnering with Product and Engineering to turn customer insight into roadmap decisions, guide complex products from concept to validation, designing practices and systems that make great design repeatable across the org.

See organizational impact →

Craft

VARIOUS

UX/UI/Design Craft →

Kind words from excellent humans

Company culture is everything. I'm proud of the strong relationships I've built with my colleagues. I help build supportive, innovative, and action-oriented teams that uplift one another and collaborate to create exceptional work. A rising tide lifts all boats.

★ Featured ★

Arkady Sokolov

AI Product Leader | ClickUp

“Yvonne is a cheetah-mounted hot pink laserbeam”

At Kustomer she killed the handoff. Not metaphorically. She moved the entire design org out of static specs and into AI-native co-creation with engineering: prototyping directly in IDEs, testing outputs against real data, shipping in hours instead of days. She codified the whole thing into reusable markdown-based skills (prompt structures, fidelity…

Jeffrey Zeldman

Executive Creative Director | Automattic

“Yvonne Doll is a one-in-a-million, multi-talented designer, artist, musician, and, above all, design leader”

It was my pleasure and privilege to watch Yvonne mentor, encourage, inspire, and support the Jetpack design team. Beyond being a model colleague, she’s also tremendous fun, supremely cool, and somehow, despite being fun and cool, a grownup and mensch. I would hire her myself if I were in a position to do so, and consider myself fortunate indeed if she…

Range | Depth

I’m a hands-on product design leader who brings executive judgment, creativity, and systems thinking to complex work. Music and abstract painting shape how I think about timing, ambiguity, restraint, and clarity, whether I’m leading a team or running the whole design show myself.

In my free time I'm an abstract oil painter and the lead singer/songwriter/guitarist for an indie rock band.

04:06

Thought Currents

All blog posts →

04 · 9/2026

Velocity.
Systems.
Impact

Leading by building

Codify Design Judgment

Make the Right Thing Easier

Check the Work, Then Investigate

Keep Judgment Human

Organizational Impact

Craft

Range w/a point of view

Thought Currents

Zero UI

AI: The death of the handoff

The Museum of Abandoned AI Features

AI-Native: Leading
by building

AI-native: Leading by building

Codify Design Judgment

Make the Right Thing Easier

Check the Work, Then Investigate

Keep Judgment Human

Codify Design Judgment

Make the Right Thing Easier

Check the Work, Then Investigate

Keep Judgment Human

Shaping product direction

Craft

Kind words from excellent humans

“Yvonne is a cheetah-mounted hot pink laserbeam”

“Yvonne Doll is a one-in-a-million, multi-talented designer, artist, musician, and, above all, design leader”

Range | Depth

Thought Currents

CROSSFADE 02: THE CLICK & THE GRID

CrossFade 01: The Abstract Painter's Product Mind

The Museum of Abandoned AI Features

Bad UI Now Has a Meter Running

Velocity.Systems.Impact

Leading by building

Codify Design Judgment

Make the Right Thing Easier

Check the Work, Then Investigate

Keep Judgment Human

Organizational Impact

Craft

Range w/a point of view

Thought Currents

Zero UI

AI: The death of the handoff

The Museum of Abandoned AI Features

AI-Native: Leading by building

AI-native: Leading by building

Codify Design Judgment

Make the Right Thing Easier

Check the Work, Then Investigate

Keep Judgment Human

Codify Design Judgment

Make the Right Thing Easier

Check the Work, Then Investigate

Keep Judgment Human

Shaping product direction

Craft

Kind words from excellent humans

“Yvonne is a cheetah-mounted hot pink laserbeam”

“Yvonne Doll is a one-in-a-million, multi-talented designer, artist, musician, and, above all, design leader”

Range | Depth

Thought Currents

CROSSFADE 02: THE CLICK & THE GRID

CrossFade 01: The Abstract Painter's Product Mind

The Museum of Abandoned AI Features

Bad UI Now Has a Meter Running

Velocity.
Systems.
Impact

AI-Native: Leading
by building