Pulpatronics — Yidan Xu

🔒 This page isn't all of the work. A good deal of what I made at PulpaTronics is still confidential, so what follows is the slice I'm free to show. More on what's missing at the end.

Where it ended up: mark construction, wordmark, tagline lockup and the rebuilt palette, as one system

Project overview

PulpaTronics makes RFID tags out of paper. No metal antenna, no chip, no laminate, which is the whole point: over 18 billion RFID tags are manufactured for single use every year, and a million items carrying a PulpaTronics tag instead saves around 2,400kg of CO₂ a year. It's a B2B sustainability play, and it was pitching to retail brands that judge a supplier partly on whether it looks like it will still exist in three years.

I joined as a design intern in London in 2024, into a team of eight, as the only designer in the company. The brief was brand building, which at that size meant whatever the brand needed that week, decided by me.

My contributions

Three strands, and each one grew out of the one before it.

Rebuilding the product's visual system, starting with colour. I audited the existing palette, found what was making it read as tired, and remapped every token in the system to a replacement. That set the constraints everything after it was made inside.
Design material support. Once the system existed, I produced against it: marketing collateral, business cards, an illustration set for a joint campaign, and visual support for customer workshops.
UI/UX design, for PulpaTronics and for the ventures it worked alongside. Interface and experience work for the company's own product surfaces and for the partner startups it ran joint work with, including the website design standards. This is the strand I can show least of, because almost all of it sits under NDA. It was also a substantial share of what I actually made there.

Rebuilding the visual system

The old palette was three colours: a deep Business Blue, a Paper White, and a Peace Blue turquoise that existed in the spec but was barely used anywhere. In practice the brand ran on desaturated earthy tones, dark blues, browns and greys, with beige and off-white doing most of the work. It read as faded paper. Appropriate for a paper product, less appropriate for a technology company asking retailers to bet on it.

The original three-colour specification: Business Blue #002632, Paper White #ECECE8 and Peace Blue #66CDDB, each with RGB, CSS and HEX values — The system as I found it: three colours, and the only bright one going unused

What the brand actually had to do

The positioning was the real constraint. PulpaTronics is a B2B startup selling an unproven material into an industry that buys on reliability, and there are two ways to get that wrong. Look too much like a startup and you lose the reliability. Look like an established supplier and you lose the only reason anyone should pick you over the incumbent.

The old palette had drifted hard into the second. Earthy, low-contrast, quiet: it read as a company that had always been there and intended to keep doing the same thing, which is the exact opposite of the claim the product makes. What it needed to read as was early and inevitable. A company on a trajectory, not one holding a position. Practically that meant keeping the deep navy, because trust is what a supplier sells, and finally spending the turquoise, because nothing in the system was saying that anything was happening.

Part of the audit, not the outcome: the mark on the backgrounds the brand was actually using at the time, warm black included. Three of the six are dark and sit at nearly the same value, so the wordmark lands with the same weight on all of them and nothing in the system can carry emphasis.

What I changed

Put the highlight colour to work. The turquoise was already in the brand book. I gave it a job: a small, deliberate accent that gives every layout a focal point.
Redefined the neutrals. Cleaner beige and off-white, so the paper reference reads as intentional rather than worn.
Balanced reliable against innovative. The deep navy stays, because trustworthiness is what a B2B supplier is selling. Brighter mid-tones carry the rest.
Aligned the palette with the mission. A blue-green range reinforces the environmental argument without anyone having to say the word "eco".

The mapping is where the work actually is. A muddy warm black came out of the system entirely. The navy was cleaned and deepened. The mid blue lifted substantially, from a near-grey rgba(40,54,65) to rgba(69,114,134). A neutral grey became an actual light blue. Pure white softened to paper white, so nothing on the page glares.

Colour system mapping: each old background, middle and font colour shown beside its replacement, with RGBA values and numbered arrows, including one colour marked Removed — Every token in the old system, mapped one-for-one to its replacement, with one removed outright

It reads better in motion than in a table, so here it is as one:

Design materials

With the system settled, the rest of the internship was output. Everything below is made inside those constraints.

A PulpaTronics business card mockup: a paper-white card reading 'YIDAN XU', 'PULPATRONICS' and '2024' in embossed type, photographed on a deep teal surface under leaf shadows

Business cards. Paper white on Business Blue, with the wordmark blind-embossed rather than printed, which is the cheapest way to make paper feel deliberate.

A slide titled 'Activity Design Support' showing eight of eighteen black-and-white line illustrations: a shopper deciding between garments, someone puzzling over a care label, a person overwhelmed by warranty and returns windows, a Petit Pli shopfront, and scenes of scanning a tag for care tips

Eighteen illustrations for the PulpaTronics × Petit Pli joint campaign, drawn to script the offline user-experience tests. Line work only, so the set could be produced fast and read at any size.

A customer workshop table: printed illustration worksheets numbered by problem, branded PulpaTronics × Petit Pli stickers, glass carafes, crisps and biscuits

The illustrations in use at a customer workshop, printed as worksheets and cut into branded stickers

Site work. The product photographed as the hero, because the tag is the argument.

Being the decision maker

PulpaTronics was eight people, and I was the only designer in it. That number changes what design is. There is no design team to sit inside, no brand guardian upstream of you, and nobody writing you a brief.

The habit a large company builds in you is waiting. You take a problem as far as your level allows, lay out two or three routes, and hand them upward for someone more senior to choose between. That's a rational habit there. It is useless at eight people, because there is nobody above you who knows more about this than you do. If the palette is wrong, no one is going to come and tell you it's wrong. You have to go and find out whether it's wrong, teach yourself whatever you're missing, and then decide.

The job stops being "present options" and becomes "own the outcome". Research it, learn what I don't know, choose, and turn up with the finished thing rather than a menu.

That last part is the half people underestimate. Deciding privately is easy. What the role actually asks is that you get people who aren't designers, and who are busy with something else, to trust a judgement they can't independently check. You earn that by arriving with a resolved answer and the reasoning behind it, not with a set of alternatives that quietly hands the decision back to them. Doing that made me the owner of the work in a way I don't think I'd have learned as one designer among thirty.

It also means the brand is never one company's problem for long. A hardware startup selling into retail lives in other people's contexts: joint campaigns, partner brands, other London startups whose own identity has to sit next to yours without either one losing. The illustration set was for a campaign run with Petit Pli, and a chunk of my remaining time went to collaboration pages and UI support where PulpaTronics appeared as somebody else's partner rather than as the host.

What that actually trained

Validating instead of escalating. At this size nobody signs off your palette. You have to establish for yourself that the old one was costing the company something, then say so in language a founder and an engineer both find reasonable.

Making the system do the work you can't. One designer cannot be in every deck, email and trade-show banner. Testing the wordmark on every background in the palette was not a nice-to-have; it was how other people made correct choices on days I wasn't in the room.

Designing at the speed the company moves. Eighteen illustrations existed because a campaign had a date. Line work only, no colour, no rendering, because that was the version that could be finished and still be good.

Working with people who don't speak design. Customer workshops, engineers, commercial leads, partner brands. Almost none of my day was spent talking to other designers, which turned out to be the most useful part.

This is the thing I'd most likely have missed if I had only ever grown up inside a large team, and it's what I took out of a startup and kept. Everything I've built since, including founding my own venture, runs on it.

The design work is only half of the job. The other half is being the person who decides, and who can say why clearly enough that other people back the decision.

🔒 Most of what I made at PulpaTronics isn't on this page. A startup's design work is its positioning before that positioning is public, so the website design standards, the later collateral and the partner work stay under NDA. What's here is the slice I'm free to show, chosen because it's the part where you can still see the reasoning. If you'd like to talk through the rest, please email me.