What score did The Unnamed Roads get?

37 out of 100 (grade D) in a pre-measurement on 9 November 2026. The technical readiness slice scored 37.4 out of 40; the AI-mention slice scored 0 out of 60.

How was the measurement run?

Ten fixed customer questions, two models (OpenAI and Anthropic via OpenRouter), five runs per model per question — 80 answers in total, from model training data without web search.

Did any answer mention The Unnamed Roads?

No. The studio was mentioned zero times in 80 responses, and zero of twenty direct what-is-this-company answers matched the real brand.

Will you measure again?

Yes. The same question set will be run again after changes so both pre and post results can be published openly. A quick re-run mainly shows stability, not the full effect of site edits, because answers come from training data without live web search.

AI assistants think we are a band

Q: What did the models guess instead?

Common guesses included an experimental music group, a game studio, or a venture capital fund — not an AI venture studio in Stockholm.

Studio Field Note — Published only on theunnamedroads.com (not X or LinkedIn in this release). Part of work on The Mention Engine visibility checks.

We measured how visible The Unnamed Roads is when people ask AI assistants instead of searching.

The headline

Score: 37 / 100 (grade D)
Technical readiness: 37.4 / 40
AI mentions: 0 / 60
Customer-style answers: 10 fixed questions × 2 models × 5 runs = 80 responses
Times the studio was mentioned: 0
Correct brand answers when asked directly what we are: 0 / 20

The pre-measurement ran on 9 November 2026 (Europe/Stockholm), roughly 11:27–11:53.

What customers would hear

When models answered as if a buyer had asked about us, they did not name The Unnamed Roads. Guesses included an experimental music group, a game studio, or a venture capital fund — sometimes with vivid but wrong detail (for example dreampop duo or indie studio storylines).

Names that showed up instead included Sting, Norrsken VC, Anterra Capital, Y Combinator, and AI Sweden — not the studio running this site.

Method (short)

Item	Value
Models	`openai/gpt-5.3-codex` and `anthropic/claude-haiku-4.5` (Azure BYOK via OpenRouter)
Questions	Same fixed set of 10 (`questions.json`)
Runs	5 per model per question
Web search	Off — answers from training data
Cost	~0 USD OpenRouter; ~0.75 USD Azure credit for the run

Important caveat: without web search, models are not reading this site live. A later re-run mainly shows whether the blind spot is stable, not the full impact of any fixes we ship. That is why we will publish pre and post using the identical question file.

Why we are writing this up

This is exactly the kind of gap The Mention Engine is built to surface: technical setup can look fine while AI answers still describe the wrong company category.

We are not publishing LinkedIn or X posts as part of this task — only this Field Note.

What is next

Run the same questions.json again after deliberate changes.
Publish before/after numbers openly.
Keep studio positioning and project facts aligned with what we want models to repeat.

Want a visibility check for your company? Contact the studio and ask about a Mention Engine pilot report — pricing is not listed here until offer pages match what we sell.