← All work

🤖 Bridge - AI exploration platform for product teams

Own product, built in public

Product StrategyProduct DesignUX ResearchDesign SystemsSAAS
Case study cover
Quick overview
PDF
Slide 1
Slide 2
Slide 3
Slide 4
Slide 5
Slide 6
Slide 7
📌
Every product team I spoke to in 2026 had started exploring with AI, and none could tell me why they kept what they kept. The exploration happened in a chat window, the good idea was copied by hand into the product, and the reasoning stayed in someone's head. Bridge closes that gap: you explore with AI in whatever tool you already use, Bridge tracks what the exploration changed, you review it, and the code, Figma kit and docs follow. I researched it, designed it, and built it.
81 sources researched before the first screen
65 AI explorations reviewed, each closed with a written reason
Live free and public on GitHub

🖥️ The product

Three roles, three different first screens. A contributor sees only what is waiting on them. A reviewer gets the queue. A lead gets adoption and drift. Nobody is shown a fact they cannot act on.

Bridge Studio home as a design lead: the people waiting on you, each row a change with its screen and its age
Bridge Studio home as a design lead: the people waiting on you, each row a change with its screen and its age

The review is the product. An exploration arrives as a screen, before and after, with the teammate's open question attached and two answers to give: needs changes, or looks right. No diff, no pull request, no code. The developers get it as it is.

The library review queue: changes to the design system itself, grouped so the grouping says the state and each row shows only what differs
The library review queue: changes to the design system itself, grouped so the grouping says the state and each row shows only what differs
The quality screen: design system adoption and drift, measured on the rendered product rather than declared
The quality screen: design system adoption and drift, measured on the rendered product rather than declared

🔍 How I framed the problem

I did not want a product built on my own frustration. Before the first screen, I built a corpus and kept it as data, one row per source, so every claim can be recounted.

What I based it onHow much
Sources in the file81
Distinct named witnesses43
Named organisations47, of which 38 are neither platforms nor research publishers
Sources at the highest evidence level61 of 81
Citable by name62 (17 anonymised, 2 private)

Where it came from matters as much as how much:

ChannelSources
Open web: Hacker News threads, opinion pieces, published studies, vendor docs38
LinkedIn posts and comments27
My own tracker: job specs received, interviews, screening calls11
Direct exchanges: video calls, private messages5

Two thirds is public and checkable. That was deliberate: a corpus a reader can verify is worth more than one they have to trust.

The research file, published in full and readable on its own
The research file, published in full and readable on its own

🩹 Five pains, and the workaround that proves each one

I only kept a pain when I could show the workaround people had already built for it. A workaround proves the pain is real and that it costs something, because somebody is already paying to route around it.

PainWhat teams do insteadWhat it costs downstreamSources
No decision is written down when an AI exploration is keptThe decision stays in a Slack thread, or in the head of whoever generated itThe same debates get replayed with no memory15
Explorations start from the design system and never come back to itWhat worked is copied by hand, or not copied at allSystem drift, falling adoption, handoff done twice31
Nobody knows who approved whatAsk the lead again, or just assumeBlurred responsibility at the moment of the bug17
The volume generated outruns human review capacitySkim the review, or merge without readingReview debt, repaid in production15
With no tool, the team builds its ownAn internal tool at an online betting operator, a hand built automation workflow at a car manufacturer, a design system written in markdownMaintaining a tool that is not their product4

The last one has the fewest sources and mattered the most. Four teams had each decided it was cheaper to build their own governance tool than to keep working without one. That is the strongest buying signal in the corpus, and the clearest statement of what the product had to beat.

The five pains in the research file, each one traceable to the sources that support it
The five pains in the research file, each one traceable to the sources that support it

🔄 What the corpus contradicted

The public conversation in 2026 was about generation: which model, which prompt, how fast. My corpus was not. Sorted by subject, 47 of the 81 sources are about what happens after something is generated: 30 on the governance of explorations, 17 on design system adoption. Eleven are about new ways of working with AI at all.

So the interesting problem was never the generating. It was the keeping. That reframing is what turned a design system tool into a governance product, and it is the most valuable thing the research produced.

🧭 Governance turned out to be a level nobody had named

I had been sorting my own decisions into three buckets by where they were stored, which classified nothing. Re-read against what each one actually commits, four levels appeared:

LevelDecisionsWhat it commits
Product14What a person sees and can do
Governance9How the work gets done, reviewed and shown. Nothing about the product
Strategy7What the product is, who it is for, what we will not build
Technical1Implementation, reversible

Nine of thirty one decisions touch no screen at all. They are about how work is reviewed and made visible, which is exactly the category Bridge sells. The taxonomy I needed to understand my own project turned out to be the product's own subject matter.

🗂️ 65 closed explorations, every one with a written reason

Bridge is built by AI agents working on branches, and every exploration ends in a closed pull request. I read all 65 and recorded why each one closed. Not one is missing a reason.

OutcomeCount
Abandoned and archived36
A test, closed on purpose11
Ran its course9
Superseded by another8
Closed by mistake1

Twenty nine of the 65 are not abandonments. Reading a closed pull request as a failure is the default, and it is wrong more than two times in five. Thirty one of the 65 were opened on a single day, as one batch decision rather than thirty one judgements, which is the kind of thing an archive hides until you count.

The exploration register: 65 closed explorations, each with its outcome and the reason it closed
The exploration register: 65 closed explorations, each with its outcome and the reason it closed

📐 The design system it runs on

Bridge runs on its own design system, around thirty components, and its interface uses only real components from it. No hand made imitations: an audit that counts an imitation as adoption is a governance tool lying about governance.

The Bridge library: foundations, components and patterns, the same pages a team gets for its own system
The Bridge library: foundations, components and patterns, the same pages a team gets for its own system
How review works, written for the people who will be reviewed, not for the people who built it
How review works, written for the people who will be reviewed, not for the people who built it

The rule I held to throughout, and the one the corpus agreed on unanimously: never take a capability away from someone already using it.

Key Result

The product is live at bridgedesignsystem.com, free to use and public on GitHub, with components and tokens under MIT and the product under a licence that turns each version MIT after two years.

The research is published in full and reads on its own: the Bridge research file, 81 sources, the five pains, the 65 explorations and the 31 decisions, each traceable to where it came from.