It Didn't Win: How a Hackathon Build Became a Working Employee Skill Exchange
In this blog
On March 13, 2026, at Love at Scale, a nine-hour hackathon run by Lovable and SwenodeAI, Shanice Brown built The Barter Shop Corporate: a platform where employees list what they can do and what they need, and AI finds the matches. It didn't place. It did produce a working prototype, and a question worth taking seriously: would an internal skill exchange actually work at WWT?
Inside any large company, there is an enormous amount of capability that never shows up on an org chart.
Someone in finance photographs weddings on weekends. Someone in engineering tutors algebra. Someone in operations has redesigned four resumes this month for friends who asked. Someone two floors up needs exactly one of those things and has no idea that person exists.
The org chart tells you who reports to whom. It tells you nothing about what people can actually do.
The Barter Shop Corporate was built to make that visible.
What it does
An employee creates a profile with two lists: services they can offer, and services they need.
That's it. Tutoring. Lawn care. Photography. Resume reviews. Technology support. Childcare. Whatever a person can genuinely do.
AI reads both lists across the whole population and surfaces the matches, including the ones nobody would think to search for. Coworkers discover each other through capability instead of proximity or title.
No cash changes hands. The exchange is the point.
The Match Studio
The marketplace is the easy part. The differentiator is the matching, and the design constraint on it is specific: it should not behave like search. It should behave like an advisor.
The AI weighs what's offered, what's needed, availability, location and reciprocity, which is the interesting part. It looks for pairs where both people get something, not one-directional requests.
Every recommendation returns three things:
A match score, 0–100. Something a person can triage at a glance.
A "why this match" explanation in plain language. Not a relevance percentage, an actual argument for why these two people should talk.
Opportunity tags. Naming what the exchange gives each person, rather than what category the service falls into.
That middle one carries the product. A search result tells you something matched. An explanation tells you why it's worth an afternoon.
The exchange loop
Join → build a profile → list what you offer → list what you need → AI analyzes compatibility → matches arrive → the two people talk → exchange proposal → exchange completed → both sides review each other → trust score rises → the network gets more useful for everyone still in it.
The last three steps do the heavy lifting. An exchange platform without a trust layer is a bulletin board. Trust here compounds from completed exchanges, ratings, reviews, repeat collaborations and transparent history, so reputation carries forward instead of resetting every time.
What's actually built
A working Lovable prototype with real surface area, not a single demo screen:
Landing page. Marketplace. Dashboard. Profiles. AI Match Studio. Exchange requests. Ratings and reviews. Messaging. Admin dashboard.
Alongside it, the system that produced it: Lovable build prompts, Magic 25 UI prompts, a design system spec, and image and animation prompt sets.
The visual direction was set as a hard constraint. Modern SaaS: matte teal, slate blue, charcoal, generous whitespace, subtle motion. Reference points are LinkedIn, Airbnb, Stripe, Mercury, Notion. Explicitly not Craigslist, not Facebook Marketplace, not classified ads.
That isn't vanity. Someone deciding whether to put their name and their skills into an internal platform reads the interface as a credibility signal before they read a word of copy.
Where this stands
Being direct about status, because it matters:
This is a prototype, not a product. It has not been reviewed, approved, adopted or piloted by WWT. No employees are using it. No data has been collected. Nothing has been deployed.
The next step is a conversation. The prototype is being brought to internal stakeholders to find out whether an internal skill exchange is something WWT would want, and what would have to be true for it to work here. Security review, HR policy, identity and access, moderation and whether people would use it at all are open questions, not solved ones.
The honest version of the pitch is a question. Not here is a platform, let's roll it out. It's here is a working prototype of an idea. Is this worth exploring?
Learn more
Because this hasn't gone through internal review, prototype access isn't open by default.
If you want to see it or follow where it goes, fill out this form: The Barter Shop Corporate: Learn More
The principle underneath
AI's job here is to make that visible.
Nine hours in March produced a build that didn't win and a prototype that still works. The next move isn't more code. It's finding out whether it belongs here.