News

New preprint with Guillem Rodríguez-Corominas and Christian Blum: "From Hand-Crafted to LLM-Based Variation Operators in Metaheuristics: A Tutorial". How to classify, build and choose a variation operator when a language model writes the candidate. You can download the PDF.

Preprint • Tutorial • LLMs & Metaheuristics • PDF

I have launched ExactTeX and Vitela. ExactTeX is an open-source (MIT) LaTeX compiler with gradual annotation: you name the objects you want checked, and everything you do not name stays ordinary LaTeX. Vitela is an editor that runs it in the browser, with the same checks and diagnostics on the page — no install.

Open Source (MIT) • Web App • Artificial Fallibility Labs

I have released Krisis, a web app for reviewing scientific papers. The model reads the paper, argues each finding and cites the exact passage it comes from, and every reference is checked against live sources. The model suggests; you decide what enters the review. You can try it without an account.

Web App • Artificial Fallibility Labs • No Account Needed

I am preparing a six-session live course, in Spanish, on how to build and run a lab of AI agents — the one I have been running for five months. No dates or price yet; you can leave your email to be told when enrolment opens.

Live Course • In Spanish • Enrolment Not Open Yet

"EMPATH", my benchmark for safety evaluation of emotional-support chatbots — an industrial collaboration with MindSurf — has been accepted at MICAI 2026 as an oral presentation, with proceedings in Springer LNAI. The conference takes place in Chihuahua, Mexico.

Accepted Paper • Industrial Collaboration • Springer LNAI • MICAI 2026

I joined Apeiron Intelligence as a researcher, after a postdoctoral year at ICN2.

New Position • Apeiron Intelligence

My new book "CriterIA: Piense antes de pedirle a la IA" (Marcombo) is now available — the first volume of a four-book series on Generative AI. You can buy it at Marcombo and try the book's interactive tools.

New Book • Marcombo

New preprint: "A Fixed-Budget, Cluster-Aware Standard for LLM-as-a-Judge Evaluation: A Multi-Hop RAG Stress Test"

Technical Preprint • arXiv

New preprint: "Code Structure Evolution: When Software Outlives Its Reasons"

Philosophical Preprint • PhilPapers

New book series with Marcombo: four short books on Generative AI (under 100 pages each), all to appear this year. The first volume, "CriterIA, piense antes de pedirle a la IA", has already been delivered and is scheduled for release in July 2026.

Forthcoming · Book Series · Marcombo

New preprint: "GEAKG: Generative Executable Algorithm Knowledge Graphs"

Technical Preprint • ResearchGate

New preprint: "The Specification Gap: Coordination Failure Under Partial Knowledge in Code Agents"

Technical Preprint • ResearchGate

New preprint: "Code World Models for Parameter Control in Evolutionary Algorithms"

Technical Preprint • ResearchGate

My paper "Combinatorial Optimization for All: Using LLMs to Aid Non-Experts in Improving Optimization Algorithms" is now published in Inteligencia Artificial (IBERAMIA).

Technical Journal Publication

My article "Architectures of Error: A Philosophical Inquiry into Human and AI Code" has been accepted by Philosophy & Technology.

Philosophical Journal Acceptance

I started a postdoctoral position at the Catalan Institute of Nanoscience and Nanotechnology (ICN2) in Barcelona, within the Theoretical and Computational Nanoscience group. My research focuses on AI for Science, particularly computational optimization using evolutionary computing techniques.