Panelia press kit
Everything you need to write about Panelia: ready-to-paste boilerplate, sourced figures, press releases, logos and a direct contact. Visuals are free for editorial use.
Last updated: September 3, 2026
About Panelia
Two lengths of standard "about" paragraph, to reuse as they are.
Panelia is an AI-powered synthetic market research platform. It simulates hundreds of respondents to measure purchase intent for a product, a price, a message or a packaging design, and returns a calibrated distribution with its confidence intervals in ten minutes.
Panelia is an AI-powered synthetic market research platform founded by Maxence Lassus. From the description of a concept — product, price, message, name or packaging — it simulates hundreds of respondents representative of a market, then returns a calibrated purchase-intent distribution with confidence intervals, verbatims and an executive summary. The method builds on research published as arXiv 2510.08338: rather than asking a language model for a rating, Panelia lets it answer freely, then scores that text by semantic similarity. Respondents are labelled as explicitly synthetic: Panelia positions itself as a fast pre-test, not as a substitute for fieldwork.
Key figures
Every value is the one shown elsewhere on the site: they cannot drift apart.
| Launch date | August 7, 2026 |
|---|---|
| Cost | −99% versus a €10,000 traditional study |
| Turnaround | 10 minutes, versus 3 to 8 weeks |
| Synthetic ↔ human correlation | 0.92 |
| Confidence interval coverage | 94% |
| Entry paid plan | from €99 per month |
| Languages | French, English, Spanish, Italian, German |
| Publisher | Maxence Lassus, sole trader (Sathonay-Camp, France) |
The correlation and interval coverage come from a versioned benchmark set (showcase-v1, 26 concepts matched against human data), published as such on the "Science" page. These are calibration measurements, not a guarantee of the result for any given concept: the reliability shown in a report depends on the size of the panel surveyed.
Scientific publication
"LLMs Reproduce Human Purchase Intent via Semantic Similarity Elicitation of Likert Ratings" — arXiv 2510.08338, October 2025. This is the methodological foundation of Panelia.
Press releases
Full texts, free to reproduce and quote.
August 7, 2026
Panelia opens calibrated synthetic market research to the teams who could never afford to commission it
Sathonay-Camp, France — 7 August 2026. Panelia is launching its AI-powered synthetic market research platform. It lets teams test a product, a price, a message, a name or a packaging design with hundreds of simulated respondents, and get back a calibrated purchase-intent distribution in about ten minutes.
The problem
A quantitative study from a research institute costs in the region of €10,000 and takes three to eight weeks. That entry ticket effectively rules out small businesses, independents and product teams who need to decide quickly. The outcome is familiar: either you launch without testing, or you test one idea out of ten.
The method
Panelia builds on research published as arXiv 2510.08338, which shows that a properly prompted and calibrated language model reproduces human purchase-intent distributions. The key point is counter-intuitive: the model is never asked for a rating. Each persona answers freely, and the resulting text is scored from the outside, by semantic similarity to anchor sentences. Asking directly for a "1 to 5 rating" produces degenerate distributions and poor correlation with humans.
Every study freezes the model, embedding, anchor and calibration versions that produced it, along with its random seed: a result stays replayable and auditable.
What Panelia does not do
Respondents are presented as explicitly synthetic, both in the platform and in exports. Panelia positions itself as a fast pre-test — to arbitrate, prioritise, eliminate — not as a replacement for fieldwork. Every report shows its confidence intervals and a reliability tier derived from panel size.
Availability
The platform is available in French, English, Spanish, Italian and German. A free trial with complimentary credits allows a first study without a payment card. Paid plans start at €99 per month.
About Panelia
Panelia is an AI-powered synthetic market research platform founded by Maxence Lassus. It simulates hundreds of respondents to measure purchase intent for a concept and returns a calibrated distribution with its confidence intervals in ten minutes.
Press contact — Maxence Lassus, founder — presse@panelia.io
Visual assets
Logos in SVG and PNG, transparent background, light and dark versions. The wordmark is converted to outlines: no font to install, no substitution risk in print.
Brand colours
- Panelia blue
#2E5BFF - Teal
#17A2C9 - Green
#1FB573 - Ink
#111111
Typefaces
Inter
Interface and body copy
SIL OFL 1.1
JetBrains Mono
Wordmark, figures and data
SIL OFL 1.1
Quotes
Ready to use, quotable without prior approval.
Serious market research used to cost €10,000 and six weeks. That price never filtered out bad ideas — it filtered out the people who could not afford to test them.
The right question is not "what score do you give this product". It is to let the respondent speak, then measure what they said. That is the whole difference between a made-up number and a calibrated one.
We do not replace fieldwork, and we say so on every report. What we make possible is testing ten ideas before taking a single one to real customers.
Press contact
Interviews, additional data, demos: reply within 48 working hours.
Maxence Lassus
Founder
presse@panelia.ioPanelia · panelia.io
Usage rights
- Logo and visuals may be reproduced freely for editorial purposes — press, blogs, publications, presentations — without prior permission.
- Please do not alter the logo's colours, proportions or spacing, and do not rebuild it with another typeface.
- The Panelia brand remains the property of its publisher: using it implies no partnership, sponsorship or endorsement.
- For any other use — advertising, merchandise, domain names — please write to us first.
Inter and JetBrains Mono are distributed by their authors under the SIL OFL 1.1 licence. Since the wordmark ships as outlines, you need no font installed to reproduce the logo faithfully.