BRENNEN

Industries / Consumer goods

Greeting cards

The greeting card trade writes, illustrates, prints and sells folded cards that people send or hand over to mark birthdays, holidays, weddings, deaths and other occasions, through card shops, supermarkets, chain stores and websites that print each card to order.

1843

In London, Henry Cole asked his friend the artist John Callcott Horsley to design a Christmas card, then had a printer make a thousand copies, each with room for a handwritten greeting. They sold at a shilling each, which the Victoria and Albert Museum (V&A) calls expensive for the time, and the venture "was judged a commercial flop." The V&A says Cole "sent the first Christmas card in 1843.".

vam.ac.uk

30

Milestones achieved across 6 domains.

8

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Henry Cole had a thousand Christmas cards printed in London in 1843, and they did not sell well. In Worcester, Massachusetts, Esther Howland began selling valentines in 1849 and filled her orders with a small assembly line of friends. In Boston, Louis Prang printed Christmas cards in color from the mid-1870s and was selling nearly five million holiday cards a year by 1880, by the account of The Henry Ford museum.

Where it stands

By the US Greeting Card Association's count, published by the US Postal Service in 2019, Americans buy about 6.5 billion cards a year, worth $7 billion to $8 billion. The UK association put its market at £1.53 billion in September 2025. Specialist chains have shrunk: Clinton Cards went into administration in 2012, and Papyrus's owner moved to close its whole chain in 2020. Card Factory, which designs and prints its own cards, had 1,117 shops in January 2026, and Moonpig, which prints each card to order, says customers used its creative features, AI-made stickers among them, in 31 million cards in the year to April 2026.

What is open

Most open jobs are tests the trade has not run on itself: whether readers prefer verses by people or by AI, whether detectors can tell AI-made card art from an artist's, which finishes stop a card being recycled, and whether people can tell robot handwriting from the real thing. Others are tools: a robot that assembles pop-up cards, a phone app that counts empty pockets on a card rack, and print software that sets messages in Arabic, Hindi or Chinese correctly. Each can be started this month with cards from a shop, open-source software and current AI models.

Domains

AchievedWatch forOpen

Advances to build with

2 research notes

Open in greeting cards

3 of 12 open deliverables

Open · Design and writing

Verses read blind

A blind reading test of card verses. For each of 40 occasions, readers see three unsigned verses: one by a professional card writer, one by an AI model, and one by an AI model then edited by a writer. They pick the one they would buy and guess who wrote each.

Built with what exists

current models such as Claude, Gemini and Mistral Large 4, whose weights are free to download (see the Note "A trillion open parameters"); verses commissioned from freelance card writers; a paid survey panel such as Prolific; a short analysis script.

Credit · unclaimed

Open · Printing and finishing

A robot that builds pop-ups

A robot cell that takes the laser-cut parts of a slice-form pop-up card, an interlocking grid of paper strips, and assembles them: slotting the pieces together, gluing the shape into the card and folding it flat.

Built with what exists

a pair of low-cost arms such as the SO-101, run with Hugging Face's LeRobot software (see the Note "No download"); a robot model that learns from a few demonstrations, such as NVIDIA's GR00T 2 (see the Note "Five demonstrations per task"), or Runway's Praxis-1, whose weights are free to download (see the Note "Open robot model weights"); a desktop laser cutter; overhead cameras.

Credit · unclaimed

Open · Shop displays

Count the empty pockets

A phone app that photographs a card rack and reports which pockets are empty, which are running low, and which hold a card in the wrong section, such as a sympathy card among the birthdays.

Built with what exists

a phone camera; an object detector such as RF-DETR or YOLO, trained in Roboflow; a small vision model fine-tuned on one graphics chip, such as the vision model in Liquid AI's LFM3 family (see the Note "One GPU to fine-tune"), to read section headers and card captions.

Credit · unclaimed