The system, called OPTIMIA, acts as a kind of “brain” that connects and interprets key information sources at the OAJ, responding by voice in real time and natural language. In essence, it is an expert AI assistant that has been under development for the past two years at the Centro de Estudios de Física del Cosmos de Aragón (CEFCA).

The project is still in an early stage but has already drawn interest from other professional observatories. This week, engineers and managers from the Gran Telescopio de Canarias (GTC) are testing it on site, and the findings will later be shared through a common platform with Spain’s main astronomical observatories, as part of a knowledge exchange initiative within the Astronomy Infrastructure Network (RIA).

OPTIMIA already performs tasks such as explaining work procedures (a useful tool for new staff), guiding users through the observatory’s main systems and documentation, transcribing meetings and drafting minutes, and sending out weekly scheduling emails. These are real examples of how this AI system is already helping to streamline and enhance internal processes. The project has a four-year roadmap, with its main goals set for completion in 2029.

By connecting systems such as maintenance and planning, the AI is designed to offer recommendations, diagnostics, and predictive alerts. It can assist with tasks as simple as forecasting the weather or identifying time windows to carry out small repairs without disrupting scientific operations.

“The idea is to help the teams work better and more efficiently — it’s not designed to replace jobs, but to support daily activities,” explains Axel Yanes, Head of Engineering at CEFCA and project lead.

An AI born in Teruel

CEFCA is the developer of OPTIMIA, but not its commercial provider. The goal is to share its design and functionalities with other astrophysical infrastructures and research facilities recognized by Spain’s Ministry of Science, Innovation and Universities as Unique Scientific and Technical Infrastructures (ICTS). This AI-based support service, developed in Teruel, is the first of its kind to be deployed at a Spanish observatory.

Developed entirely by CEFCA’s own teams, the system combines several forms of neural-network-based Artificial Intelligence, integrating Large Language Models (LLMs) with deep learning and machine learning techniques — models that learn and build structured layers of knowledge.

“You can think of these systems as different team members that assist with daily tasks,” notes Yanes.

Potential applications range from database access and information management to process monitoring and incident resolution. OPTIMIA can simplify authentication and security procedures and, most importantly, support real-time troubleshooting.

“It gathers data from the observatory’s control system, internal databases, and ongoing tasks. If a technician faces an unfamiliar situation, OPTIMIA can instantly suggest possible steps or procedures, saving valuable time,” adds Yanes.

Yanes also emphasizes CEFCA’s commitment to keeping the system under human supervision at all times:

“OPTIMIA will never operate autonomously, nor transmit or store sensitive data. It is designed exclusively to assist personnel on site,” he explains.

The system complies with European, Spanish, and Aragonese regulations on Artificial Intelligence (HAP/795/2025), ensuring a transparent and ethical development process.