We build projects in artificial intelligence and technology, applied across different fields of knowledge.
Our projects:
Chat intake → One-Pager
The agent asks the right scoping questions for your data, your domain and your goal. Risk and legal bases are processed inside the EU AI Act + GDPR framework from the first turn.
Project Plaza → Collaboration Match
Your One-Pager is visible to verified AI Researchers in the Project Plaza. They apply at your tiers — research collaboration, MSc, BSc, classroom. You review and accept the match.
Project Room → Publication
The collaboration unfolds in the Project Room — a shared workspace with downloadable audit trail. Output: a paper, a tool, a methodology — published in the open-access Journal.
Trust is not a section in our terms of service. It is the foundation every conversation, every match, every export sits on.
01
EU data residency
Your data does not leave the European Union — at rest, in motion, or in transit through model inference.
02
Zero retention on LLM processing
All LLM and embedding providers operate under no-train, no-retain agreements.
03
EU AI Act · Limited risk
Scheria is classified as limited-risk under Article 50. Each One-Pager carries its own classification with article-level citations.
04
Auditable by design
Per-project audit log, downloadable, ready for institutional DPA review.
Calidoscopio: AI to learn better, not to stop thinking.
In-person training on the responsible use of artificial intelligence. We bring the rigor of a master's in Responsible AI to classrooms, community centers and communities, translated into practice —with judgment, safety and without outsourcing thought.
In-person courses in small groups, close and free of jargon. Critical thinking about AI: verify, cross-check and understand. Tailored programs backed by local councils and institutions.
Crisopeya · Calidoscopio
Crisopeya · Project 03
Theia
A geometric morphometrics lab, entirely on your phone.
Capture, automatic landmark digitization and full statistical analysis in the field —entirely on-device, with no PC, no cloud and no internet.
- Edge AITwo native YOLOv8 models detect the flower and place 32 landmarks on the specimen. Runs on Android phones under €100.
- Human in the loopModel confidence is mapped to a color spectrum; the researcher adjusts by hand and keeps final authority. It is not a black box.
- Full analysis, offlineOn-device GPA + PCA: interactive morphospace, deformation wireframes and a pair of reproducible FAIR files.
Case study: Erysimum · BioChange, University of Granada
Manuscript under review at Methods in Ecology and Evolution.
Digitization · 32 AI landmarks