Synthetic data. Every entry below was generated by a large language model from biographical scaffolding. This is platform demonstration — not empirical evidence about any real population.
Demonstration cohort

What the platform produces, side by side.

Twenty synthetic personas — all LatAm→Spain — written into existence in two parallel cohorts. Ten wrote diary entries punctuated by Prometheus’s AI-driven guided questions; ten wrote diary entries alone. Same corridor, same time window, same biographical depth. The register differs. That difference is the point.

§ 01 Disclaimer

What this is, and is not.

This dataset comprises entirely synthetic narratives generated by large language models (OpenAI gpt-4o-mini for subject voice, Anthropic Claude Sonnet for comparative analysis) from biographical scaffolding. The 20 personas were designed for breadth (national origin, indigenous and Afro-Latino identities, queer and trans identities, working-class through professional, secular and religious) but they do not represent real people.

It is intended exclusively for demonstration of platform capability and dogfooding of the analytical pipeline that will be applied to real elicited dialogue. It is not empirical evidence about any population. Any apparent resemblance to real individuals or communities is coincidental and unintended.

The methodological evidence that does carry empirical weight is documented in the manuscript Eliciting the Narrative Register (Vasse, 2026), which uses real elicited dialogue alongside the Blog Authorship Corpus as a negative control.

The two excerpts below are from two different synthetic personas in two different conditions. They are illustrative of the platform’s intended output character, not a controlled comparison. The contrast is hand-selected for clarity.
§ 02 A two-line illustration

How the platform’s intended outputs differ in register.

One persona, in dialogue with Prometheus, the AI assistant, answering "where do you feel most at home." Another persona, writing a diary entry alone, on an unremarkable afternoon. Both LatAm→Spain.

persona_01 — Argentine queer software developer, Madrid

Elicited · Q1

Home. It’s a complicated word, isn’t it? I’ve felt it tugging at me since we moved to Madrid, but I’m not sure it’s stuck yet. Sometimes I catch myself in the narrow streets of Lavapiés, surrounded by the shouts of vendors at the rastro, and I feel a spark of something that resembles belonging. Like yesterday, watching Camila negotiate for a vintage dress while our dog, an overexcited little ball of fur, tried to steal the show by barking at every passerby. But then there are moments when that warmth evaporates, and I remember the distance—both physical and emotional. The Slack messages from my coworkers blaring through my headphones, the frustration of trying to translate everything into a different cultural context. The tech talk feels alien sometimes, a jargon that runs circles around…

persona_17 — Paraguayan Guaraní-speaking mother, Madrid suburb

Unelicited · diary

hoy no fue un día fácil. la casa estaba un poco desordenada, pero mariana y diego estaban contentos, jugando con sus juguetes. les hice un chocolate caliente en la tarde. a ellos les gusta, siempre piden más. después, hicieron ruido y corrían por todo el piso; a veces me asustan y tengo que recordar que son pequeños. llamé a mi madre como cada semana. su voz es un abrazo, pero me duele saber que ella está enferma. me dijo que ya no puede salir tanto y eso me preocupa. “aichejáranga, mami,” le dije. le prometí que vendré a visitarla, aunque no sé cuándo. el cielo estaba nublado, y el frío llegó temprano. la luz del día se apagó rápido. me siento sola en esta ciudad tan grande, sin amigas cerca. me gustaría tener a alguien con quien compartir el café en la tarde. pero miro a mis hijos y…

The platform is designed to produce the introspective register on the left and to distinguish it from registers like the diary entry on the right. Whether elicited dialogue with real participants will reliably exhibit this contrast — and how large the effect will be — is the empirical question the platform exists to investigate.

§ 03 By the numbers

What sits in the cohort.

20

synthetic personas · 10 elicited + 10 unelicited.

832

total entries · arcs of 14–36 months between 2022 and 2025.

$0.36

total cost · gpt-4o-mini generation + Claude Sonnet comparative analysis.

Read the methods note

Claude Sonnet’s comparative analysis.

An 8-section markdown methods note covering register comparison, theme distribution, voice diversity, what the demonstration shows, what it does not show, and explicit DOL-pipeline expectations.