On 22–23 June 2026 I presented at HEALTHY INDOORS in Valladolid, the final scientific conference of K-HEALTHinAIR (Horizon Europe grant 101057693). There is no recording. The session was Session 4, Data Analytics, Artificial Intelligence and Standardised Methodologies for IAQ Assessment. I am the sole author, for inBiot Monitoring SL, Pamplona.
Content
The talk presented Anne, an AI assistant I am building at inBiot to read indoor air quality data from sensors and assess it against standards such as WELL, GO IAQS and the EU buildings directive (EPBD). If you paste sensor data into a general chatbot, the chatbot does the arithmetic itself: it can invent thresholds, and it can return a different score each time you ask.
Anne is built so the language model does not do the calculations. The air quality scores, statistics and charts come from separate software tools that the assistant calls, so the same sensor data always gives the same result. A person reviews the findings before Anne suggests any action. The abstract reports three things: a consultant can see which tools ran and on what data, the numbers are tested for consistency, and the assistant is already in early use, from a primary-school classroom in Budapest to a live dashboard at an event in Brussels. A formal study with several independent assessors is still to be done.
Publication
The book of abstracts came out later.
- Printed title: Architectural guardrails for AI-assisted IAQ assessment: An agent reasoning across WELL, GO IAQS, EPBD and beyond
- Book: Healthy Indoor Congress. Frontiers Abstract Book (Frontiers Media SA)
- Page: 244
- Editors: Alicia Armentia (University of Valladolid), Jose Fermoso (CARTIF), Sandra Rodríguez-Sufuentes (CARTIF)
- ISBN: 978-2-8325-8036-3
- Cite the book as: Armentia, A., Fermoso, J., Rodríguez-Sufuentes, S. (2026) Healthy Indoor Congress, Valladolid, Spain. June 22–23, 2026
- License: The abstracts were not Frontiers peer review; each is CC-BY 4.0