Pierce.

Building Products · London

Pierce Paeste

I build products in London, shipped fast and demoed live. My edge is a fashion-business background: I don't just build the thing, I know how to position it and sell it.

See the work Get in touch

About

From fashion school to the demo stage

I studied Fashion Management at London College of Fashion. Building came later, through London's hackathon scene. I turned up curious, shipped something real with people I'd just met, and got hooked.

Now that's what I do: AI products, built fast, demoed live, made for the person on the other end rather than for the tech itself.

The fashion degree wasn't a detour. It's the edge. Brand, sales and marketing came first for me, so I don't stop at a working demo: I position it, tell its story and sell it. Plenty of builders can build. Fewer can sell what they build.

Projects

Four hackathons. Four shipped products.

Every one went from idea to live demo inside the event, two hours at the fastest and a week at the longest.

Cursor iOS Hackathon · London 3rd Place · 200+ Builders Built in ~2 hrs

RoomCast

An AI interior designer in your pocket.

Furnishing a room means guessing what fits, what matches and where to buy it. Scan a room with your phone and RoomCast suggests furniture to suit it, in a range of styles, with buy links pulled from multiple marketplaces. We built it in about two hours, entirely from the Cursor iOS app on our phones, with no laptops.

3rdof 200+ builders
Top 5live demo on stage
~2 hrsbuilt entirely on phones

Role First hackathon for the whole team: me, Johan and Jethro Repato, all of us building from our phones.

  • Cursor
  • SwiftUI
  • Apple RoomPlan
  • Claude
  • Built on a phone
OpenAI Build Week · London 1-Day Build

SlopGuessr · AI or Not?

Wordle for spotting AI.

AI literacy is becoming a basic skill, and almost nobody has practised it. One short daily set for everyone: text, voice and video, some human-made, some AI-generated. You call each one Real, AI or Not sure. The moment you answer you see the proof and the tell, so you don't just get a score, you get better. A live leaderboard and a Kahoot-style head-to-head duel run on the same daily set.

Role One of a team of four with Yu Hang Joachim Sin, Abdulhamit Celik and Harry Parker, taking it from premise to playable in a single day.

  • OpenAI
  • Daily game
  • Team of four
Fleek × a16z · London 1-Day Build

DROP

A clothing bale in, a priced and ready-to-list collection out in minutes, not days.

Resellers buy secondhand clothing by the bale, then spend days grading, pricing and listing each piece by hand. Point your camera at the pile: DROP grades each piece, prices it with reasoning you can read using price bands rather than fake precision, writes the listing and exports to eBay or your own storefront. In one day, the two of us shipped live scanning, bulk photo import, price banding, a curated storefront, CSV export and eBay integration.

Role One half of a two-person build with Johan Repato, including the call to rebuild the core concept mid-day after the Fleek team showed us what buyers actually need.

  • Vision pipeline
  • Two-person team
  • eBay integration
  • CSV export
AMD Developer Hackathon: ACT II 1-Week Build

Iolite

Runna does running. Iolite does everything else.

Multi-sport athletes train for real events across running, swimming, strength and more, but most training apps only speak one sport. An AI coach that builds event-anchored, periodised plans across eight disciplines, with sessions backed by a real 3,000-exercise library, natural-language rescheduling and a calendar that handles multiple events at once. Built on the AMD AI stack via Fireworks AI.

Role One of four on team Hackadawgs: me, Johan, Jan and Jethro Repato, building across the full week.

  • Fireworks AI
  • lablab.ai
  • 3,000-exercise library
  • Team Hackadawgs

Skills

What I actually do

No skill bars, no invented years of experience. Three things, each proven the same way: by shipping.

AI product building

  • Rapid prototyping: idea to working demo inside a day, with Cursor, Claude, OpenAI and Fireworks AI.
  • Prompt design: the working core of every build, from RoomCast's room suggestions to DROP's price reasoning.
  • Structured outputs: price bands, listings and training plans a product can rely on, not fake precision.
  • Vision pipelines: point a camera at a room or a clothing bale and turn what it sees into product.

Shipping under pressure

  • Scoping to the demo: cut to what can be shown live before the event ends.
  • Building with strangers: with whoever's in the room, from a two-person build to a team of four.
  • Live demos: called up out of nowhere at my first event and demoed in the top five.
  • Four-for-four record: two-hour, one-day and week-long builds, every one shipped, one on the podium.

Brand, sales & marketing

  • Positioning: the fashion-business edge. I don't stop at a working demo, I position it and sell it.
  • Storytelling: brand, sales and marketing came first for me; every build ships with its story.
  • Sales: real shop-floor experience at a London fashion retailer: in-store, online and at pop-ups.
  • Pitching: pop-ups, e-commerce and the demo stage. Selling the thing, not just building it.

Background

Where the edge comes from

First builds

Self-taught e-commerce. Built and ran my own Shopify storefronts and dropshipping ventures before AI was the arena. Learned what makes people buy by doing it, not studying it.

Education

Fashion Management, London College of Fashion (UAL)

Then I found hackathons, and it turned out fashion school is great training for shipping.

Currently

What has my attention

Two builds and one scene.

Building

GEO consultancy

Helping businesses get found by AI search, showing up in the answers, not just the links.

Building

Pitch

A sports venue finder for London, built on Next.js and Supabase.

In the room

London's AI builder scene

Cursor, OpenAI, Claude Code and King's AI events, plus the next hackathon with my team.

Connect

Say hello

WhatsApp is the fastest way to reach me. Everything else below works too.