In our "Learn with Our Experts" series, we invite specialists to #ShareViews on current trends in education. Today's guest writer is Marcos Blanco, who has been with Santillana for over 13 years as a content creator and editorial coordinator. With a specialization in environmental ecology and expertise in learning technologies, Marcos combines his passion for natural sciences, technology, and AI, focusing on developing digital educational content. In this essay, he shares an interesting use case of AI in Santillana's content creation workflow, reflecting on the main benefits, challenges, and lessons learned.
GenAI in content creation: Santillana’s case
By Marcos Blanco, AI local lead at Santillana
The intersection between artificial intelligence and education is no longer a futuristic concept, it's a present-day reality. At Santillana, we are actively exploring how generative AI (GenAI) can support and enhance our editorial processes. Our workflow is overseen by human experts who make the final decisions, while our teams strictly follow ethical AI guidelines. The goal isn’t to replace human creativity, but to enhance it, making content creation more agile, accessible, and adaptable. Over the past months, we have put this vision into practice in two key areas: multilingual translation and content generation.
Accelerating translation: 80% faster with AI
Let’s start with translation, an essential but historically time-consuming part of educational publishing in Spain, where four different main languages coexist with Spanish (or Castilian), the official language throughout the whole country.
Using an AI-powered platform, we’ve been able to cut translation times by over 80% in some projects. For instance, translating 27 secondary education notebooks into Catalan, Galician, and Basque previously took up to 60 days. With the help of AI, the same process now takes just about a week.
This leap in efficiency is not just about speed. It's also about flexibility. While translations required revision, particularly for linguistic accuracy and cultural fit, the ability to run AI translations in parallel across multiple teams freed up valuable time and resources. We can now revise and improve our Spanish-language originals for longer, without compromising delivery timelines in other languages. In other words, AI has helped us move faster and raise our quality standards at the same time.
Creating multilevel content with AI assistance
On the content generation side, we’re experimenting with AI to create multilevel learning activities, especially for subjects like Math, Biology, and Physics. Our teams prompt AI tools to generate exercises of varying complexity, aligned to curricular goals and structured templates. The tools often provide helpful starting points: complete with titles, problem-solving steps, and even solution keys, reducing the editorial workload by up to 60% in some cases. For example, in some of our science notebooks for secondary education, AI-generated content included accurate calculations, coherent tables, and curriculum-aligned activities that teachers could immediately adapt.
But let me be clear: this isn’t a push-button magic solution. Vague prompts produce inconsistent or overly simplistic results and AI-generated outputs still need careful editorial review: the human aspect is crucial for developing our trustworthy, high-quality content. Without oversight, Gen-AI tools may confuse content level or miss curriculum nuances, and certain tasks like creating visual elements, accurate maps, or complex equation, require human intervention. However, they significantly reduce the time it takes to draft, allowing editors to focus more on refining and less on starting from scratch.
Lessons learned (so far)
Through these pilots, we’ve learned some important things:
1. Speed is just the beginning: Yes, AI helps us move faster, but its real value lies in enabling better allocation of time, allowing teams to focus on creativity, pedagogy, and quality control.
2. Structure matters: AI works best when content inputs follow clear templates. The more structured our prompts and formats, the better the results.
3. Human review is non-negotiable: While some content types (like reinforcement tasks) are well-suited to automation, others such as diagrams, lesson plans, or specialized curriculum adaptations, still require deep human input. Across all use cases, from translation to activity generation, expert revision ensures that what AI produces meets our educational standards. AI can accelerate drafting, but expert validation is essential for educational integrity.
4. Language models aren’t always visual thinkers: When it comes to diagrams, maps, or science schematics, we still need hands-on design and editorial intervention.
Where do we go from here?
At Santillana, AI isn’t replacing what we do, it’s helping us do it better. It allows our editorial teams to iterate faster, our translators and proofreaders to work more efficiently, and at the end, our customers to access richer content with less delay.
As we continue refining these tools and exploring others across Sanoma, our focus remains clear: leveraging technology to empower learning. AI is not an end, it’s a means to deliver more accessible, high-quality, and adaptable education.
We're just at the beginning of this journey, and I believe the best way forward is through open dialogue. To all fellow editors, and innovators: let’s share experiences and build a collective understanding of how AI can serve our shared goal: better learning for every student. I’d love to hear your thoughts. How are you using AI in your context?
This essay is part of a series of texts about the topic to be published in June. Stay tuned to our channels to find out more about AI in education.