"Is your taste in this?"
ARTIFACT · early 2025, as it actually was
A real Figma file or review board from then, cropped to nothing internal. Needs export.
Docker, early 2025. Nine designers, each embedded in a product team. Most weren't doing much with AI.
"We're not in the time of optimizing how to use Photoshop. The rules are changing every week."

No mandate, just different fronts

The agentic product surface · [name]

A design system built to be read by agents · [name]

Fast trials: Figma Make, Claude Code, v0 · [names]

I asked the three or four who were already excited to each take a different front.
ARTIFACT · the room
One weekly show-and-tell, someone's screen mid-demo. Doesn't exist yet (and it can't be the "brain" vault).
Every week. What worked, what didn't. Designers who struggled got a peer to learn from, not pressure.

I chose not to chase every engineering ticket to prevent slop. Policing all of it would burn out my designers.

ARTIFACT · GitHub as the design workspace
A cropped repo view of the agent-ready design system, built with Claude Code. Needs export, nothing unreleased.
About a year later: the whole team works mainly in agent tools and barely opens Figma. Two design engineers joined.
"We changed how the organization approached product work. It's not about catching up. It's about shifting the way design works so that the impact of design remains true."
Still open: engineering moves so fast that some designers end up playing catch-up. Consistency across teams isn't solved.
ARTIFACT · Nobody uses the front door
A Google results page → one click → you land mid-Hub on an image page. The homepage never appears.
Needs export (a screen recording or two stills).
ARTIFACT · today's Hub · needs export
"The entire catalog of everything"

Six rounds

1
2
3
4
5
6
One frame per round, from Figma. [confirm all six exist] · live: hub-vision.vercel.app/prototype
ARTIFACT · Sean's published Hub Vision piece, byline visible
[link / location]
"I want my designers to believe and know they could do better."
Design: Sean Brady. It's shaping the Hub roadmap; nothing has shipped yet.
Should Facebook decide what's true?
2017 · News Feed, 1B+ monthly users · Product design: Cat Audi and Earl Carlson, with 4 engineers, 1 researcher, 1 content strategist, 1 PM · [title, end date]
No.
25 fact-checker partners, 14 countries.
"We wouldn't have had a tool that was designed so that fact checkers could do their job."
A "disputed" flag on a Feed story
2023 deck, slides 14–25 [check it's there]
For the people most primed to believe it, the warning read as proof.
Label or demote was a group decision, made on evidence. I pushed for demotion; so did others.

~80%

drop in views once an article was rated as containing misinformation

The front of the shelf

Publisher logos
needs export
About This Publisher
needs export
Trust Project indicators
ethics · corrections · ownership · masthead
Credibility signals and publisher logos were my push. Cat Audi and the team led much of this craft.
faint: India research page · 2023 deck · needs export
"I helped take what the research team was doing and condensed it down into something that was actually actionable."
"Snopes isn't covering stories in Brazil or India."

A journalist told us Facebook wasn't safe to post on. Cat and I pushed for tools for that.

"People were getting a lot of their news in Facebook group posts and WhatsApp groups…"
"How do we start to work on WhatsApp groups…"