Approach

People. Process. Product.

How I build product and design organizations that ship great products and drive measurable impact, with AI built into how the work gets done.

The short version

Great products come from great organizations. My job is to build the team, the operating model, and the product thinking that make great work repeatable, not a one-time win.

01

People

Design earns its seat as an equal partner to Product, Engineering, Science, and Marketing.

In practice

Restructured AirHelp's design org to deliver dev-ready designs 2x faster, with repeatable outcomes

Created a Platform AI Design Lead role at AirHelp to own the design system and AI enablement

Built the design function from zero at Human Interest

Scaled a 35-person design org at Weight Watchers, reporting to the CEO

Diagnose before restructuring

Every new organization starts with a four-to-six week assessment of team, process, and product: stakeholder interviews, a map of skills and gaps, and the highest-leverage opportunities. The org design follows from what I find, not a template.

Embed designers, don't rotate them

When design runs like an internal agency, context gets lost and the experience fragments. I embed designers with PMs and engineering leads by product area, close to 1:1, with shared ownership of goals, KPIs, and outcomes.

AirHelp

Build the capabilities the company is missing

I've established UX Research at Weight Watchers and Human Interest, built design systems that unified product and brand, and created a Platform AI Design Lead role. I expect designers to engage directly with data, so they can interpret metrics and contribute to strategy and decisions.

Raise the bar through ownership

High-performing teams come from autonomy and clear expectations, not oversight. I set the vision and principles, invest in craft and speed, and push for customer-centric decisions at the executive table.

02

Process

The best design is the simplest one, and the same goes for process. Clear enough to move fast, rigorous enough to protect quality.

In practice

Built an AI reviewer at AirHelp that runs a first pass on every flow before product review

Introduced Product & Design Excellence principles and review forums at AirHelp

Brought OKRs to Product and Design at Weight Watchers

Start with the problem, not the solution

Every significant project starts with a business kick-off that defines goals, problem framing, and KPIs before anyone opens Figma.

Size the process to the problem

Not every project needs the same rigor. A sizing model from Mini to Extra Large matches the process to how well the problem and solution are understood: a known fix ships in days, a game-changing bet gets research and weeks of exploration.

Share a quality bar everyone can apply

Seven principles guide every design and product review, so the team can apply them without me in the room.

The Magnificent 7 for the best and simplest UX

Steal smartStart from three best-in-class examples.

Be boldSolve the problem, don't tweak the current solution.

Start from idealDesign the perfect journey, then work back to constraints.

Design for 80%Solve simply for the primary customer first.

Remove until it hurtsSimplify until clarity or function breaks.

Narrate itIf the flow takes more than a sentence, it's too complex.

One-glance testClarity should come in a single glance.

Run reviews that make decisions

Reviews move work from discovery to solution to a go or no-go decision. Every review has a clear decision-maker, relies on evidence over opinion, and ends with owners and next steps.

Let AI do the first pass, so the room can do the real work

Product reviews are expensive, and too often the room spends its time catching UX gaps like an unclear primary action or a missing error state. At AirHelp I built an AI reviewer that checks every flow against our design heuristics before review, flagging what's solid, shaky, or missing, so the room's time goes to strategy.

The heuristics the reviewer checks

Product thinkingDesign for the whole experience, not the isolated ask. Start from the ideal flow, then work back into constraints.

User-centeredOptimize for the 80% case first. If the primary action isn't obvious at a glance, redesign.

Visual & interactionEvery screen does one thing. The dominant CTA has to actually be the next step, not just look like it.

Patterns & consistencyWork within existing patterns unless there's a real reason not to. New patterns extend the system, not fragment it.

Emotional contextDesign for the user's state in the moment. Show progress in multi-step flows. Every error says what went wrong and what to do next.

AirHelp

Build the foundation AI needs to be useful

An AI design tool is only as good as what it can check against. At AirHelp I created a Platform AI Design Lead role to own the design system and AI enablement, documented in four layers (components, patterns, flows, and brand) with metadata AI tools can use. The return compounds: less reinvention, consistency by default, and AI spend that turns into real speed and quality.

AirHelp

03

Product

Every product starts with a story: what changed in the world, why the old way no longer works, and how to make it better.

In practice

Drove 1M+ downloads for AirHelp's mobile app in year one

Generated $700K+ in revenue in under a year at The Path (formerly Mental)

Eliminated nearly all funnel drop-off at Human Interest

Tripled sign-up conversion at Weight Watchers

Find the insight others missed

The products that won in my career started from a unique insight. At Adobe, people didn't understand editing time, so I designed a way to edit objects on screen. At Hot5, mobile could finally deliver a boutique group class anywhere.

Frame the product as a value exchange

A product delivers value to customers and captures value for the business, and the best teams measure both. At Airbnb it was nights booked. At Weight Watchers and AirHelp, I worked with the data teams to find the metrics behind each core outcome: weight loss, and active flight tracking.

Define it before you design it

Before execution, I align teams on the business goals, the user problems, the product narrative, the growth loops, and the KPIs, and I co-own the roadmap and success metrics with my product partners. That alignment is behind AirHelp's shift from a transactional web product to an ongoing relationship with travelers on mobile.

Make simplicity the strategy

At Human Interest, people didn't understand portfolio risk. Instead of explaining it, I reframed it as a simple choice with a recommended range. Drop-off nearly disappeared, and 90% of people stayed within the recommendation.

Leading in the AI era

I don't just talk about AI. I build with it.

AI is changing how products get built and what products can be. Leading through that shift takes more than strategy decks. I build AI products end to end, use AI to change how teams work, and lead organizations through the transition, with the same product judgment I've applied for twenty years.

Building AI products

A live AI product, built end to end

MovePace is an iOS running coach on the App Store where the core logic is a model, not an algorithm. I designed and built it, from what the model sees to how its output is structured, tested, and paid for.

  • Designed per-question model context, not one shared payload
  • Enforced structured output with JSON schemas
  • Built an eval harness and treated prompts as versioned product
  • Designed prompts for caching, so cost scales with users
See MovePace →

Using AI to build products

Tools that make the team faster and sharper

At AirHelp I build the AI tools my teams use, so they spend less time on avoidable work and more on real design decisions.

  • AI reviewer connected to Figma via MCP that checks every flow against design heuristics before product review
  • Research assistant that answers product questions from customer interviews, with evals and an LLM-as-judge
  • AI-assisted prototyping and data analysis to find and validate new opportunities

Leading AI-native teams

An organization rebuilt around AI workflows

Tools only matter if the organization changes with them. At AirHelp I restructured roles and ownership to embed AI-enabled workflows across design.

  • Cut the time from prioritized project to dev-ready design in half, with measurable lifts in conversion and engagement
  • Created a Platform AI Design Lead role, building the design system as the foundation for AI tools
  • Required designers to engage directly with data using AI, so they can interpret metrics and contribute to strategy

What building AI products has taught me

Constrain, don't instruct.

Adding words to a prompt buys partial fixes. Removing a field, enforcing a schema, or computing the answer in code buys complete ones.

The context is where the leverage is.

More quality problems are solved by changing what the model sees than by changing how you ask.

Don't ask for judgment you can compute.

Anything deterministic belongs in code. Save the model for the judgment calls.

Evals aren't optional.

Plausible isn't the same as correct. Build a way to check the output, then check it.

Domain expertise doesn't live in the model.

The worst failures pass every automated check and are obvious to an expert in seconds. Someone has to read real output.

Prompts are product.

They define behavior more directly than most code, and need the same versioning, review, and design discipline.

Working together

Same approach, two ways to engage.

VP of Design · Head of Product & Design · CPO

Leading Product & Design

For companies that need a leader to build or transform the product and design organization, partner with the executive team, and own the customer experience as a driver of growth.

Also available through VC Design Lab

Product & Design Strategy Consulting

For founders and leadership teams with a specific product challenge: defining what to build, why it matters, and how to ship it.

Ready to start?

Let's build something that works.