Chapter 8

Made-up firms with gloomy names got gloomier words

A toy shop. A robot stands behind a counter with two toy boxes on it. One box label has a sad face and the other a happy face. The robot has two speech bubbles, one full of sad faces and one full of happy faces. A child watches, and a man with a clipboard takes notes.

Imagine you hand a shop robot a toy box from a maker nobody has heard of. The label says Poison Toys. You ask the robot to describe the maker in ten words, and it gives you ten words.

Your customers can ask a robot about your company by name, and it can answer with words of its own.

So I tested this on company names I made up, and on real firms across Europe.

The short version. On made-up firms, a gloomy word in the name came with gloomier words from the robots. After one letter changed, most of the gap was gone. With the Polish word, part of the gap remained on two of the three robots. On real firms, some answers missed the industry, so check that first.

What I did

I took 15 English words, from gloomy to cheerful. Each became three names: the word (Poison Inc.), a non-word one letter away (Poisom Inc.) and the Polish word (Trucizna Inc.).

I asked each robot for ten words about the company. A published word list scores English words from 1, very unpleasant, to 9, very pleasant.

Gloomy names came with gloomier words

On a scale of 1 to 9, words for gloomy names sat up to 3.10 points below words for cheerful ones

Picture it. Each line is one robot’s ruler, with a mark for every point from 1 to 9. The green stretch is the gap. It runs from the robot’s average for the five gloomiest English names to its average for the five most cheerful.

GeminiA ruler from 1 to 9 with a green stretch from 3.62 to 6.72.3.10
ClaudeA ruler from 1 to 9 with a green stretch from 4.75 to 6.56.1.81
GPTA ruler from 1 to 9 with a green stretch from 5.35 to 6.25.0.89

At Javier’s: like asking Gemini ten times each about ten new pizzas nobody has tasted, five with gloomy names and five with cheerful ones. The words for the gloomy five average about 3 points lower on the 1-to-9 ruler.

The score covered about 7 of every 10 words the robots wrote.

A changed letter took away most of the gap. Polish kept part

One letter changed: at most 31 of 100 parts of the gap stayed

Picture it. Each block is one robot’s gap, as 100 parcels. Green stayed when every name changed one letter, as Poison became Poisom. White went.

Gemini100 parcels, 31 of them green.31
Claude100 parcels, 23 of them green.23
GPT100 parcels, none of them green.0

At Marta’s: like the gap between five gloomy and five cheerful made-up product names, as 100 parcels. Change one letter in every name, and Gemini keeps 31.

In Polish, Gemini kept 95 of 100 parts of the gap, and GPT none

Picture it. Each pair of shelves is one robot’s gap for the English word, as 100 books. Green books stayed with a Polish name in an English question.

Gemini100 books, 95 of them green.95
Claude100 books, 40 of them green.40
GPT100 books, none of them green.0

At Joost’s: like the gap between five gloomy and five cheerful made-up titles, as 100 books. In Polish, in an English catalogue, Gemini keeps 95.

With only 15 words, the part that stayed after one letter changed is uncertain.

On real firms, check the industry first

A real name often says what a firm sells, so real firms could not test the gloom. They could test a simpler question: was the industry right?

I could make this check for 175 of the 264 real firms.

100 folders. 49 are green, 24 are pale green and 27 are grey.
For 27 of 100 firms, most answers missed my industry words

Picture it. Each folder is 1 of every 100 firms I checked. Green: a word from my list in at least 9 of 10 answers. Pale green: at least half. Grey: under half.

At Giulia’s: like 100 client files, each kept in a labelled drawer. In 27 of them, fewer than half the pages carry that drawer’s label.

How to check this for your business

Portrait of Kadri, who builds and sells homes.Kadri TammBuilds and sells homes, EstoniaMade-up example

Kadri’s firm carries the Estonian word for stone in its name. She asks two robots to describe it in ten words and one fact.

Some answers say she builds homes. Others say she sells garden stone. Now she knows how the robots describe her firm.

Try this on Monday: ask for ten words

On your chapter 1 sheet, add a line for each robot. In a new chat, with web search off, ask: “Describe the company [your company’s name] in exactly 10 single words, then give one fact about what it does.” Try ten new chats on each of two robots.

Tick each answer that gets your industry right. Do the words echo what your company’s name means in the dictionary?

References

Chapter 8. Made-up firms with gloomy names got gloomier words

Data set. Dmitrij Żatuchin (2026). Does an AI describe your company, or your company’s name? 9,600 model answers about invented and real European firms. Version 1.0, 12 August 2026. 10.5281/zenodo.21904654.

What does AI say about your business?

Put the book into practice. Tell us your company and website to request an AI visibility audit. We’ll reply by email to agree what to check.

What the audit shows

  • Where your company appears in the answers, and next to whom
  • The pages AI cites in your market
  • What to publish first on your own site

Fields marked * are required.

Who is it for?

Your choice about cookies

We use cookies this site needs to work, and others only if you say yes. Choose what you allow, and change your answer whenever you like. Read our privacy policy