What if a T-shirt could work as a small dictionary?
That was more or less the starting point for Pictiq, a project I have been working on around a very simple idea: sometimes you do not need to translate a conversation. You just need to communicate one thing quickly.
Water. A taxi. A toilet. Wi-Fi. Can I pay by card? Where is this place?
We already use visual languages like this every day. Road signs, airport pictograms, maps and public symbols work precisely because they do not try to express everything.
I wanted to see what happens if you approach the same idea systematically.
So Pictiq gradually stopped being a collection of icons and became a small visual protocol.
Every concept is a framed tile. Pointing at a tile is already a valid message. A few operators add questions, urgency, negation and quantity. Short sequences create simple phrases.
water + ?
taxi + card + ?
medical + !
The system is intentionally constrained. Most messages should fit into one to three tiles, and five is the limit. Pictiq is not supposed to replace natural language. The interesting question is how much useful communication can happen before natural language is necessary.
Context changes the vocabulary
The next experiment was to make Pictiq contextual.
For Paris, for example, I created a small set containing recognizable city-specific symbols alongside reusable concepts. The Eiffel Tower, the Louvre pyramid and Notre-Dame can coexist with food, bars, parks and the universal travel vocabulary.
That produced a Paris T-shirt where the larger tiles represent the local context and the smaller ones handle universal needs.
The same principle works on much smaller objects.
For a nightlife experiment, a disposable lighter carries only three symbols on each side: smoking, cannabis and drinks on one side; love, safer sex and beer on the other.
The object itself provides part of the context, so the vocabulary can become smaller.
That became one of the principles I like most about the project: do not ask how many icons can fit. Ask which icons earn the space.
From pictures to a protocol
Underneath the graphics, Pictiq is deliberately machine-readable.
Every tile has a stable ID. There is a Core lexicon, context profiles, reusable layouts and canonical SVG assets.
A person sees three pictograms.
Software might see:
[need_water, qty_2, punct_question]
That creates some interesting possibilities around computer vision, LLMs, translation, physical interfaces and personalized communication tools. A camera could eventually recognize a Pictiq sequence. An LLM could translate between natural language and tile IDs. The same message could exist on a shirt, in JSON or on a screen.
I do not know yet which of these directions will prove useful. That is part of why I built the project.
The first handbook
I have now collected the first version of the system into a free illustrated handbook:
Pictiq: A Visual Protocol for Short Messages
It covers the protocol itself, the Core lexicon, the Paris context experiment, shirts and wallet cards, portable layouts, and some deliberately less practical experiments, including translating a scene from Alice’s Adventures in Wonderland into Pictiq and seeing what happens when you try to compress poetry into tiles. The handbook also looks briefly at computer vision and language models.
The whole project is on GitHub. The repository contains the protocol, icons, packs, layouts and source material, and the handbook can be downloaded there as PDF or EPUB.
This is still an experiment. Next I want to see what happens when Pictiq moves further into physical objects, merchandise, books and perhaps software.
For now, though, the basic idea remains very small:
A tile is a message. Pointing can be enough.