May Fanucci · Principal Product Designer

May Fanucci

Principal Product Designer

For years I built B2B software. Now I design for AI. I vibe code, so I'm , but the real work is : how people and AI should fit together, without losing the human in it. The patterns are still being written, and I want to help get it right.

Enterprise AI · Document intelligence

Bottleneck by Design

87% faster document processing through collaborative AI for global banks and professional-services firms.

Visuals on request

87%
Faster document processing: 75%
Reduction in errors: 93%
User satisfaction score

The calls I made

Media · Broadcaster, via consultancy

Revenue Ceiling

2.5x ROI by transforming spreadsheet chaos into digital partnership.

Visuals on request

2.5x
First-year ROI against build cost: 89%
Efficiency gain to self-service: 98%
Fewer booking errors

The calls I made

Logistics · Same-day delivery

Solo at Scale

Dual design systems and a 47% efficiency boost across 10K+ logistics accounts.

Visuals on request

47%
Development efficiency increase: 40%
User engagement increase: 30%
Faster feature delivery

The calls I made

The screens are classified while this work is live. Full write-ups and visuals are available on request, just get in touch.

Kind words

“May was working with AI and prototyping with it well before it became a common topic in design. She's genuinely AI-fluent. And she moves comfortably across areas that usually sit with product management: structuring problems, documenting clearly, framing complex topics so the team can act on them.”

Sergio Amarante

Design Manager, Teya

“May's impact on the quality of our interfaces and user experience was huge. She translated user research into clear design changes, followed all the way through to delivery. I'd recommend her to anyone looking for a high-impact Design Lead.”

James Knight

COO, Gophr

“She led product design for a design system used across multiple teams. She inherited a challenging Figma landscape and methodically rationalised, consolidated, and refined it into a clear, maintainable system, working closely with engineering to define shared design tokens.”

Ian Lovell

Delivery Lead, Consultancy

How I work

Frame
01
Reject the brief, find the real problem
The brief is usually a stakeholder's solution wearing a problem's clothes, so I reframe around where the difficulty actually lives, often the system behind the screen. AI widens the aperture and compresses a week of research into an afternoon, but the read I keep is rarely the obvious one it offers first.
Explore
02
Widen the options, then narrow on purpose
I generate broadly to react against, grounded in how real products handle the pattern rather than letting AI invent from nothing, and I choose the interaction model the task needs instead of defaulting to a chat box. Volume isn't progress, so I bin the plausible-but-generic and keep the direction that earns a second look.
Decide
03
Commit, and own the call
I name the tradeoff out loud: what got sacrificed, why this over the alternatives, what I'd block under pressure. AI structures the tradeoff space and stress-tests the direction, but a fluent rationale for a weak call is still a weak call, so the decision stays mine.
Build
04
Make it real, at a bar that holds
This is where I vibe code: describe the interface in plain language, let AI generate working code, then rework it by hand against what good looks like, because the first prototype is always wrong and taste is the part that doesn't survive automation. For anything other people rely on, I back the taste check with a simple eval so it holds at scale.
Validate
05
Test with the people it's for
I define what success means before testing, then read the result for the signal that changes direction, not the one that confirms what I hoped. AI synthesises hundreds of sessions no one could sit through and plays critique partner before a client sees it, but it flattens nuance toward the average, so I dig past the tidy summary.
Ship & learn
06
Ship it, then find what was true
I decide what done means and what to watch once it's live, reading what comes back for the signal that should change the next cycle rather than the noise that flatters the last one. AI keeps the documentation and market view live; the learning seeds the next frame, which is how the loop closes.