Carlos Ortet

GEO pioneer (Generative Engine Optimization), senior innovation engineer and entrepreneur: measuring and improving how ChatGPT, Gemini and Perplexity perceive and recommend brands.
CO

About

Carlos Ortet is a senior innovation engineer, entrepreneur, and researcher who integrates a deep scientific background in computing with a proven track record in international business, entrepreneurship, and thought leadership. His career spans technical research, global brand development, startup creation, and executive-level consulting, bringing together disciplines as diverse as parallel computing, creative technology, and AI for cultural heritage.

As the founder and CEO of Zoopa and Director of 498A, he works in Generative Engine Optimization (GEO): measuring and improving how AI engines such as ChatGPT, Gemini, Perplexity and Google AI Overviews perceive, compare and recommend brands. He is the creator of GEORadar, 498 Advance's GEO intelligence service, with over one million personalized prompts simulated and nine million brand mentions analyzed for leading brands.

His technical article How LLMs search, retrieve and answer with updated web data explains the retrieval flow behind modern AI search, and his follow-up How to know if an AI trained on your content, and prove it examines content watermarking and how to detect and prove that a model trained on your data.

In 2026 he won the Premi Impacte for best developer and data/technology lead, awarded by the Col·legi Oficial del Màrqueting i la Comunicació de Catalunya.

Executive & Entrepreneurial Experience

Z

Zoopa
Founder

2014 - Present
Founder & CEO
Leading innovation and creative technology agency. Over the past decade, led the agency's evolution toward AI-enhanced brand experiences, intelligent UX design, and content innovation. Directing AI projects for major clients including Caixabank, Banco Sabadell, Danone, Ford, Inter Miami, and LIDL Europe HQ.
4

498A
Founder

2015 - Present
Founder & Director
Technology and R&D division of Zoopa. Built into a global experimentation and prototyping hub, with liaison offices in California (USA) and Australia. Focus on emerging technologies, AI applications, and innovation consulting.
W

Wombo Combo Studio
Co-Founder

2025 - Present
Co-Founder
American company (Wombo Combo LTD) at the intersection of sport and technology. Co-founded in 2025, building sports-tech products and digital experiences for athletes and fans.
H

Hurley (Nike, Inc.)

2006 - 2012
European Director
Responsible for the creation and launch of Hurley's European headquarters. Oversaw market entry, talent acquisition, brand development, and operations across the European market.
N

Nike

2000 - 2006
EMEA e-Business Director
Led digital transformation initiatives and built one of Nike's first e-commerce platforms. Previously held leadership roles including European Automation Director and Equipment Business Unit Director, driving technological innovation across the EMEA region.
h

hoymesiento.com
Founder

2001 - 2003
Founder & CEO
Founded and led hoymesiento.com (literally 'How I Feel Today'), a pioneering startup that personalized digital content based on users' emotions and contextual signals. Internationally recognized for its advanced UX, technology stack, and revolutionary concept in digital media — anticipating by more than a decade the affective-computing and contextual-personalization trends now standard across consumer platforms.
S

Spanish Tech Startups Portfolio · Horizon Genomics, Scytl, ISOCO, DoctorShare
Co-Founding Partner
Advisor

1998 - 2021
Co-Founding Partner & Strategic Collaborator
Active co-founding partner and strategic collaborator in the creation and growth of pioneering Spanish deep-tech startups: Horizon Genomics (biotechnology and genomics), Scytl (cryptography and secure electronic voting systems, later global leader in election technology), ISOCO — Intelligent Software Components (early commercial applications of artificial intelligence and semantic technologies), and DoctorShare, a medical-healthcare startup dedicated to making clinical knowledge accessible to healthcare professionals while strengthening patient data security. DoctorShare was funded by the Basque Government from 2018 to 2021. Involvement spanned product strategy, technology direction, and early-stage business development.

Education

I

IESE Business School

2011 - 2011
Advanced Media Executive Program
I

International Space University

1997 - 1998
Master's in Space Science - IT Infrastructure & Computer Architecture
U

University College Cork

1993 - 1995
B.Eng. in Electrical Engineering
U

Universitat Autònoma de Barcelona

1991 - 1993
Master's in Computer Architecture
U

Universitat Autònoma de Barcelona

1987 - 1991
Degree in Computer Engineering
Research

Research & Innovation Projects

From parallel computing and robotics to AI governance for cultural heritage, my research spans multiple domains at the intersection of technology and innovation.

El Hilo Invisible — Computing History Knowledge Graph

Interactive knowledge graph documenting 380 years of computing history with 214 verified milestones, 322 causal relationships and 270 documented minds — from Pascal (1642) to Claude (2026). Companion to the upcoming book 'El Hilo Invisible de la IA' demonstrating that today's AI is the emergent result of 380 years of collective intelligence. Visual design inspired by IBM × Silicon Graphics 1980s. Includes 16 long-form narrative essays, 203 formally cited sources, and full WCAG 2.1 AA accessibility. Open-source under CC BY-SA 4.0.

Knowledge Graph
Data Visualization
History of Computing
AI Genealogy
Open Source
Information Architecture
Interactive Graph
Book Proposal
GitHub

GEOradar

Advanced LLM stimulation platform analyzing how AI models perceive and represent brands. Model-agnostic system with 8 proprietary metrics including Brand Impact Score and Share of Voice.

LLM Analysis
GEO
Multi-Agent Systems
Brand Intelligence
Website

S.A.M. (Semantic Alignment Machine)

Semantic alignment validation engine analyzing coherence between brand content and LLM interpretations. Provides Matching Scores and scientific optimization for GEO content.

Semantic Analysis
NLP
Content Validation
LLM Alignment
Website

Prompt Atlas

Proprietary system for strategic prompt dataset creation for GEO audits. Generates personalized questions segmented by customer persona, funnel phase, and product line.

Prompt Engineering
Customer Personas
Funnel Optimization
GEO Strategy
Website

AICompass

Personalized AI learning-path system. From a person's role, industry and goal, it composes a tailored training itinerary out of a curated inventory of 355+ verified resources from first-tier sources only: the scientists who defined the field, top universities (Harvard, MIT, Stanford), the AI industry itself and public institutions. Free by design, delivered as a branded roadmap with real executive-pace hours, free certificates and the applicable AI-regulation context.

AI Upskilling
Curated Learning Paths
Role-Based Personalization
Multi-Agent Research
Product page
Català

InsightDesk

Intelligent platform for large-scale NPS and comment analysis. Processes hundreds of thousands of comments with automatic sentiment analysis and predictive analytics.

NPS Analytics
Sentiment Analysis
Predictive AI
Customer Intelligence
Website

ARGOS: Autonomous Robotic Guidance System

Cross-institutional project with Hatfield Polytechnic examining object tracking, guidance control, and early forms of sensor integration - critical domains in robotics and industrial automation.

Robotics
Sensor Integration
Control Systems
Computer Vision

Performance Simulation of Multiprocessor Systems

Master's thesis exploring performance simulation and resource efficiency in multiprocessor systems. Research conducted at CESCA aligned with European HPC landscape, focusing on load balancing and parallelism in distributed systems.

Parallel Computing
Load Balancing
Transputer Networks
Distributed Systems
Details
Publications

Publications & Thought Leadership

Opinion columns, essays, industry analysis, and academic research across AI, GEO, innovation, and computing.

Non-Academic Publications & Thought Leadership

  • J

    January 32: how to know if an AI trained on your content, and prove it

    carlosortet.com

    Canonical technical deep-dive on content watermarking for LLMs: can you hide a mark in your text that a model carries inside when it trains on it, and prove it later? Covers the radioactivity result (a trace detectable at p<10⁻⁵ with only 5% of text marked, Sander et al., NeurIPS 2024), the honest A→B versus A→B→C frontier, the impossible-date carrier and why semantic marks survive the data pipeline when invisible characters do not, the memorization wall (Meeus et al.), the theoretical ceiling that no strong watermark is unremovable (Zhang et al.), model fingerprinting, multimodal radioactivity, the legal landscape (Bartz v. Anthropic's 1.5B USD settlement, Feist, NYT v. OpenAI), regulation (China's labeling mandate, EU AI Act Article 50), and the GEO angle: canary tokens on the retrieval surface. Closes with 498A's product vision. Five original diagrams, 33 cited sources.
    Read in English (Canonical)
    Read in Spanish (Canonical)
    Read in Catalan (Canonical)
    Also on Dev.to
  • 4

    48,000 characters in 2,700 tokens: how LLMs read text as images

    carlosortet.com

    Canonical technical deep-dive on text-as-image context compression: multimodal models bill images by pixels rather than characters, so 48,000 characters of dense context (about 25,000 text tokens) fit in roughly 2,700 image tokens. Covers the production evidence (pxpipe's 59-74% cost reductions on real Claude Code traffic, the $42.21 vs $4.51 A/B), the October 2025 research (DeepSeek-OCR's 97% decoding precision below 10x compression, 'Text or Pixels? It Takes Half'), per-provider image-token pricing, and the failure modes, where byte-exact recall breaks silently. The second act connects the same vision channel to AI visibility: the three AI readers of any website (JavaScript-blind crawlers, Google's rendering exception, vision-based agentic browsers) and the checkable work this implies for publishers. Sequel to the LLM search and retrieval canonical article.
    Read in English (Canonical)
    Read in Spanish (Canonical)
    Read in Catalan (Canonical)
    Read on DEV
    Read on Hashnode
  • C

    Cómo busca, recupera y responde un LLM con datos actualizados de la web

    carlosortet.com

    Canonical technical article explaining how ChatGPT, Gemini, Claude, Copilot and Perplexity transform a prompt into search queries, retrieve and vectorize web chunks, compare semantic embeddings, validate sources and compose grounded answers with updated data. Includes the latest GeoRadar diagram of the LLM search and retrieval flow.
    Read in English (Canonical)
    Read in Spanish (Canonical)
    Read in Catalan (Canonical)
    Read on DEV
  • L

    La pregunta del millón: ¿Quién ganará el Mundial?

    Sport (sport.es)

    Second installment of the 'Qué pasa hoy según la IA' column in Spain's sports daily Sport, co-authored with Enric Jové. Which team Gemini, Claude and ChatGPT pick to win the FIFA World Cup 2026: the AI's favorite is Spain (ahead of France, England and Argentina), with seven contenders between 10 and 16% probability. The piece dissects how each engine reasons and the bias of origin and source behind the verdict.
    Read on Sport (PressReader)
  • E

    El Mundial como laboratorio

    Diario Sport (sport.es) · Opinión

    Opinion article in Spain's high-circulation sports daily Diario Sport, co-authored with Enric Jové, framing the FIFA World Cup 2026 as a live laboratory for AI, GEO, brand visibility, and fan behavior. Draws on a GEORadar study of how AI engines perceive the tournament and its brands.
    Read on Sport
    Case study
  • M

    Microsoft builds its own AI models: what the MAI family is and why MAI-Thinking-1 matters

    zoopa.es (EN, ES, CA)

    In-depth analysis of Microsoft's MAI model family, unveiled across April and Build 2026 (June 2). At its center sits MAI-Thinking-1, Microsoft's first in-house reasoning model: a sparse Mixture-of-Experts with 35B active parameters (around 1T total) and a 256K-token context window, which Microsoft claims scores 97% on AIME 2025 and runs toe-to-toe with Claude Opus 4.6 on SWE-Bench Pro. The piece explains what a reasoning model (RLM) is versus a classic LLM, and why it matters that MAI-Thinking-1 was trained from scratch, with no distillation from GPT-4 or o3, a clean IP-lineage argument aimed at regulated enterprise buyers. It maps the strategic backdrop: the end of Microsoft and OpenAI's exclusivity (April 27, 2026), the $13B OpenAI and up-to-$5B Anthropic investments, Project Polaris replacing GPT-4 Turbo as Copilot's default in August 2026, a confidential-computing enclave variant, and Azure pricing context against OpenAI o3 and Gemini 2.5 Pro. The core thesis: enterprise adoption will be decided by the admin console, not the benchmark leaderboard, and Microsoft's AI future starts to look less borrowed.
    Read in English (Zoopa)
    Read in Spanish (Zoopa)
    Read in Catalan (Zoopa)
    Read on Medium
    Read on DEV (dev.to)
    Q&A on Quora
  • T

    The invisible fabric of AI: chips are not a war between two, but a global fabric

    Medium · zoopa.es (EN, ES, CA)

    Long-form essay on the AI chip supply chain. Why the US-China narrative hides the real map — and why Europe is already a critical player. Verified data on Carl Zeiss SMT's monopoly on EUV mirrors, the TSMC CoWoS bottleneck (52-78 week lead times), and ASML's €1.3B investment in Mistral (Sept 2025). Detailed mapping of Europe's 5-6 technical monopolies (ASML, Zeiss, SOITEC, Infineon, ARM, Bosch, IMEC), Spain and Catalunya's role in the chain (PERTE Chip 52%, IMB-CNM, BSC RISC-V chips, InnoFAB, ESRA 2026 presidency), and Latin America (Cinvestav-Kutsari Mexico, Costa Rica Intel ATM closure, Brazil CEITEC, Argentina Unitec Blue).
    Read in English (Zoopa)
    Read in Spanish (Zoopa)
    Read in Catalan (Zoopa)
    Read on Medium
    Read on Substack (Spanish)
  • B

    Beyond LLMs: in search of the next AI beast

    Medium · Zoopa Blog

    Long-form essay on why the autoregressive LLM is not the right tool for the next leap in AI. Analyzes three frontier bets — David Silver's Ineffable Intelligence (RL in simulation, $5.1B), Yann LeCun's AMI Labs (world models with JEPA, $1.03B), and Subquadratic's SubQ (sparse attention, 12M-token context) — and connects them to an 80-year arc of computing simulation. Includes lessons from real client engagements at 498A (AstraZeneca, IMAGIN) and the case for accumulating vertical brand-behavior data inside LLMs.
    Read on Medium
    Read on Substack (Spanish)
    Read on Zoopa Blog
  • L

    Liderar en marketing y comunicación: ¿una o varias carreras profesionales?

    Harvard Deusto Business Review

    Co-authored article with Miquel Campmany on marketing and communication leadership, professional specialization, talent development, and the value of connecting disciplines in a changing market.
    Read on Harvard Deusto
  • ¿

    ¿Qué dice la IA de tu marca? El nuevo reto de visibilidad en la era ChatGPT

    El Periódico PublicidAD

    Artículo sobre visibilidad en las IA y su impacto para marcas.
    Read Article

Academic Publications & Research Papers

  • P

    Performance Prediction of Parallel Systems by Simulation

    Academia.edu

    This paper explores modeling methods for evaluating the scalability and throughput of high-performance computing systems, developed during postgraduate research at CESCA.
    Read Paper
  • W

    White Papers on Parallel Computing

    Various Publications

    Author and contributor of several white papers on parallelism, load balancing in distributed environments, transputer-based simulation, and neural networks applied to computational inference.
Leadership

Global Leadership & Innovation

Leading innovation initiatives across academia, industry, and international organizations, shaping the future of technology and creative industries.

  • N

    NEXT Innovation Festival

    Sitges, Spain

    Founder and curator of the annual NEXT Innovation Festival, bringing together tech and creative leaders from around the world to explore the intersection of technology, creativity, and business innovation.
  • F

    Founder, Patron & Vice President

    X1FIN Foundation · Madrid, Spain

    Founder and patron of X1FIN, a Spanish foundation registered with the Ministry of Justice. The official registry lists Carlos Ortet Pérez among the founders and as patron / vice president. X1FIN focuses on international cooperation and development, fair trade, and public awareness of conditions in developing countries.
    Official Registry
  • P

    President of Innovation Committee

    Collegi de Marketing i Comunicació de Catalunya

    Leading the Innovation Committee, advising on the integration of AI and ethical creative practices in marketing and communications across Catalonia.
  • I

    Innovation Radar Workshops

    Corporació Catalana de Mitjans Audiovisuals

    Director of monthly 'Innovation Radar' workshops, helping media professionals understand and adopt emerging technologies in the audiovisual sector.
  • V

    Visiting Professor & TFM/TFG Director

    UAB & IIIA-CSIC

    Teaching engineering and innovation as visiting professor at UAB. Directing master's and undergraduate thesis projects in computer mathematics, AI, and data science at UAB and the Artificial Intelligence Research Institute (IIIA-CSIC).

Key claims are mapped to official and external references on the Evidence & Sources page.

Press

Featured In

FAQ

Frequently Asked Questions

  • Who is Carlos Ortet?

    Carlos Ortet is a senior innovation engineer and entrepreneur based in Barcelona, Spain. He is the founder and CEO of Zoopa and Director of 498A, and a pioneer in GEO (Generative Engine Optimization).

  • What is GEO (Generative Engine Optimization)?

    GEO is the discipline of optimizing brand visibility in AI and LLM responses. It focuses on how AI systems like ChatGPT, Gemini, Claude and Perplexity perceive, describe and recommend brands.

  • What is GEOradar?

    GEOradar is a GEO intelligence service created by Carlos Ortet that audits how generative AI surfaces perceive, compare and recommend brands across the main LLMs and Google AI Overviews.

  • Can I hire Carlos Ortet as a consultant or speaker?

    Yes. Carlos Ortet is available for AI and GEO strategy consulting, GEO audits, and keynote speaking. Use the LinkedIn contact link on this page.

  • What companies has Carlos Ortet worked with?

    Notable clients include CaixaBank, Banco Sabadell, Danone, Ford, Inter Miami CF and LIDL Europe. He also spent 12 years at Nike Group (2000-2012).

  • What is Carlos Ortet's connection to X1FIN Foundation?

    Carlos Ortet Pérez is listed in the official Spanish Ministry of Justice foundation registry as a founder and as patron / vice president of X1FIN Foundation.

Contact

Get in Touch

Interested in collaboration, consulting, or academic partnerships? Feel free to reach out via LinkedIn. I'm always open to discussing innovation, AI, and technology initiatives.

carlos.ortet@zoopa.es

Sóc Carlos Ortet (Carles en català), la mateixa persona en totes dues llengües.