Domain-adapted LLM systems
We design domain-adapted LLM systems for regulated sectors and high-complexity environments, with curated knowledge, expert validation, and a focus on traceability.
A Universitat de València spin-off
Your technology partner for building and deploying AI
We design, validate and deploy AI systems for regulated sectors, public institutions and organisations running complex technical operations — with the traceability and oversight those settings require.
Backed by the Universitat de València and the CSIC, and grounded in published research.
Backed by
Spin-off of the Universitat de València
Approved as the university’s first AI spin-off, April 2026.
Methodology co-owned with the CSIC
Developed at IFIC, the Institute for Corpuscular Physics (UV–CSIC), and protected by IP rights.
Based at the Parc Científic
Universitat de València science park, Paterna.
What We Do
INFERA Labs builds tailored LLM systems and specialized AI workflows for regulated sectors, public institutions, and organisations operating complex technical systems. Our portfolio combines AI governance, anomaly detection, predictive modelling, multimodal analysis, expert validation, and practical training in the effective use of AI assistants and AI agents in deployment-ready environments.
Key Capabilities
Seven areas where we take technically demanding problems into reliable, deployable systems.
Governance and reliability for AI systems
We help structure AI systems that are monitorable and auditable, ready for governance frameworks, risk documentation, and evolving regulation.
Anomaly and rare-event detection
We develop methods to detect rare events, operational deviations, and anomalous behaviour in multivariate, temporal, or sensor data.
Predictive intelligence for complex environments
We apply predictive models, simulation, and uncertainty-aware analysis to networks, infrastructures, and complex dynamic systems.
Expert validation in the deployment loop
We integrate experts into the design, tuning, and validation loop to ensure operational relevance and reliability in real deployments.
Multimodal AI for technical settings
We build solutions for scientific data, signals, sensors, and multimodal integration in technically demanding settings.
AI agent training
Practical training in the effective use of AI assistants and AI agents for organisations at any level of AI maturity. We offer hands-on formats ranging from introductory sessions to advanced workshops tailored to technical, administrative, research or management teams.
Where we work
Four areas where we have already built and published working systems. Each one links to the research behind it.
Public administration & smart city
Real-time environmental and traffic alerting built on the sensor networks a city already has — forecasting, thresholds, and alerts operators can act on.
Deployed in Valencia · Neural Computing and Applications, 2025
Discuss a municipal pilotCritical infrastructure & early warning
Forecasting systems that quantify their own uncertainty, so operators know not only what is predicted but how far to trust it.
Space Weather, 2023 · geomagnetic storm forecasting
Discuss an early-warning systemHealth & medical technology
Machine learning on instrument and sensor data — event selection, reconstruction and image quality in physically demanding measurement settings.
Radiation Physics and Chemistry, 2025 · Compton camera imaging
Discuss a clinical or instrument projectRegulated industries & compliance
Language models that read legislation, standards and technical guidance and turn them into obligations, steps and deadlines — each traceable to its source.
Methodology co-owned with the CSIC · ML on Spanish parliamentary corpora, 2026
Discuss a compliance workflowHow we start
Feasibility assessment
A fixed-scope engagement that answers one question: is this problem solvable with AI, on your data, to a standard you could actually deploy? You get a written assessment — what is feasible, what it would take, what could go wrong, and whether it is worth building at all.
Scope and price agreed before we start. If the answer is no, you get that in writing too.
Request an assessment- 1UnderstandFeasibility, data, constraints
- 2DevelopBuild against real workflows
- 3ValidateExpert review and testing
- 4ScaleDeploy with monitoring
Each step is agreed only once the previous one has produced results.
Technology approach
We combine expert-curated knowledge, iterative model adaptation, and validation grounded in real use cases to develop specialized AI systems for demanding technical and institutional contexts.
Our approach prioritizes operational reliability, traceability, monitoring, and continuous refinement in response to evolving technical, regulatory, and organizational requirements.
Curated knowledge and domain context
Expert-curated knowledge and domain context shape system scope, data selection, and practical model behavior from the outset.
Iterative adaptation and continuous evaluation
Models are adapted, tested, and reassessed against real workflows so performance remains relevant under operational conditions.
Oversight and operational robustness
We design for supervision, monitoring, and resilient deployment where reliability and accountability matter.
Governance and trusted deployment
AI systems in regulated environments require more than technical performance. They must be auditable, monitorable, and capable of fitting into governance, risk management, and continuous oversight frameworks.
Readiness for AI governance frameworks
Support for internal governance structures, implementation pathways, and control models suited to high-stakes deployments.
Risk documentation and evidence trails
Documentation practices, evidence trails, and accountability records aligned with responsible deployment needs.
Monitoring and drift detection
Operational monitoring approaches to track behaviour, surface drift, and maintain model performance over time.
Support for ISO/IEC 42001 readiness
Preparation-oriented support for teams aligning AI processes, records, and controls with ISO/IEC 42001 expectations.
A traceable, specialised AI
Every claim links to the article or official register it came from, and to the date of that data. It marks whether an obligation already binds you, whether it will bind you on a set date, or whether it is still only a bill. And when a question needs human judgement it stops instead of answering for the sake of it. This is a demonstration, not an advisory service.
The answers are computed in advance from Regulation (EU) 2024/1689 as consolidated on 27 July 2026, from the BOE and from Spain's official plant protection register. None has been validated by an expert yet, and the demonstration says so.
Talk to us about a system like this for your sectorThe research behind the work
Our scientific direction rests on published, peer-reviewed research. Verónica Sanz, our Chief Scientific Advisor, is Full Professor of Theoretical Physics at the Universitat de València and a researcher at IFIC (UV–CSIC).
One figure from each paper, all co-authored by Verónica Sanz. Sources: arXiv:2007.14462, 2309.02010, 2203.03669, 2103.06115, 1906.11282, and the open preprint of the parliamentary-speech study. The painting shown is in the public domain.
See publications and metricsFounders
INFERA Labs brings together scientific depth, technical judgment and project-building experience in advanced AI.
Jaime Alcalá
CEO & Co-Founder
Jaime Alcalá is CEO and Co-Founder of INFERA Labs. Trained as a physicist at the University of Valencia and holding an MBA from Universidad Nebrija, he works at the intersection of scientific thinking, innovation strategy and project development. His experience includes the structuring of complex technical initiatives, the preparation of competitive proposals and the development of partnerships across research, technology and innovation environments. At INFERA Labs, he focuses on translating technically demanding ideas into viable projects, collaborations and deployment pathways, helping connect advanced AI capabilities with concrete sectoral needs.
Verónica Sanz
Chief Scientific Advisor & Co-Founder
Verónica Sanz is Chief Scientific Advisor and Co-Founder of INFERA Labs. She is Full Professor of Theoretical Physics at the University of Valencia and researcher at the Institute for Corpuscular Physics (IFIC, UV-CSIC). Her work combines scientific modelling, data analysis and artificial intelligence, with expertise spanning particle physics beyond the Standard Model, effective field theory, dark matter, axions and advanced AI methods for complex, high-impact problems. At INFERA Labs, she provides the scientific and technical vision behind the company’s approach to domain-adapted AI, anomaly detection, complex-system intelligence and robust decision-support tools for high-stakes environments.
Training in AI assistants and AI agents
We offer practical training programmes for organisations that want to adopt AI tools effectively, from introductory sessions to more advanced, workflow-oriented formats. Suitable for teams at any level of prior experience.
Contact
Get in touch to discuss tailored AI solutions, technical collaborations, or training for your organisation.
We also offer practical training in AI assistants and AI agents, from introductory sessions to advanced workshops for teams at any level of experience.