The challenge no one asked for

Education already had plenty on its plate. Staffing pressures, initiative fatigue, workload, accountability demands. These are not new. And then, generative AI arrived, and suddenly everyone is expected to have a strategy.

EdTech companies feel they have to integrate it. Schools feel they have to adopt it. Universities feel they have to prepare people to work with it. At every level, there’s a new kind of pressure, and a new kind of uncertainty, about how to respond in a way that is both meaningful and responsible.

The noise is real

What we see in response is understandable: a flurry of pilots, product integrations, and internal strategy groups. But the result is often fragmentation. People of good faith, working in isolation, with limited sharing of what works, what doesn’t, or why.

We’re also seeing the very real risk of surface-level use cases. A chatbot here, a writing prompt there. They’re easy to implement, but what has actually been gained? Or lost? In the midst of so much noise, it’s necessary to pause and ask what educational value is really being delivered.

Going back to the basics

One of the things the EDT&Partners team has come back to time and time again, is the idea of foundations. What if, instead of starting with the newest tool, we began by asking what schools, universities, and EdTech providers actually as a basis upon which to build well?

That is what we have tried to do with Lecture. It is not a product or a platform to adopt. It is a set of modular, open-source tools that form an evolving framework. A framework that can be used as a foundation for building AI responsibly in education. It includes things like translation, lesson planning support, exam preparation scaffolds, and feedback tooling—but those are just starting points. What sets it apart is that it is designed specifically for education, with safeguarding, pedagogy, and classroom realities in mind. We are also helping institutions put in place governance structures, risk protocols, and both pedagogical and technical guardrails to ensure AI is introduced with confidence and clarity.

Where we see an opportunity

We see two challenges currently at play. The first is longstanding: improving teaching and learning, reducing administrative burden, and creating more equitable, effective learning environments. Generative AI, used thoughtfully, can support this work by freeing up time, surfacing patterns in data, and enabling more adaptive learning.

The second challenge has been introduced by AI itself. Right now, many education organisations — schools, universities, and EdTechs — are trying to build very similar things, often at the same time. AI tutors. Planning assistants. Feedback tools. Some of it is excellent. Most is rushed. And quite a bit is costing time and money that would be better spent elsewhere.

That’s part of what we’re trying to address with Lecture. It offers modular components and supports flexible use of different LLM models depending on the task. As EDT&Partners founder Pablo Langa often says, You don’t need to bring a cannonball to swat a fly. Our aim is to help others create right-sized solutions, without having to build everything from scratch. That can mean freeing up teacher time by easing planning or feedback. It might mean enabling scaffolded critical thinking exercises that respond to where a student is in their understanding. Or simply making everyday processes, like translation, curriculum mapping, or documentation, much less of a burden.

A foundation built together

Lecture is being developed in partnership with colleagues across schools, universities, and education companies. It is designed to run within an institution’s own infrastructure, with no vendor lock-in. This gives organisations full control over how it is deployed, how data is managed, and how it connects with existing systems.

Lecture has taken shape through early-stage collaborations, supported by AWS. One example is the University of Luxembourg, where the university integrated Lecture into its existing learning platform, enabling instructors to generate auto-graded assessments, scaffold feedback, and design personalised learning content. Faculty were involved from the start through co-creation workshops and responsible AI training, ensuring the technology served pedagogical goals and respected academic standards.

Not AI for teachers. AI with teachers.

We want to move beyond simple one-to-one interactions: a teacher using an AI tool, a student working with an AI tutor. These use cases aren’t bad, but they are limited.

Instead, we’re exploring more dynamic approaches, where teacher judgment, student progress, and system design inform each other. Lecture is being used to test this in practice: from co-designed lessons and scaffolded exam prep to structured feedback and adaptive materials. These are not just ways to save time: they’re opportunities to deepen teaching and learning.

This isn’t about adopting our tool

Lecture is one step toward a shared foundation of generative AI for education. We're continuing to shape it with new partners, including a current collaboration with the University of Bath. We're committed to sharing what we learn along the way, so together, we can build faster, go further and share the lessons learned in the process.

As we continue to navigate these questions, we want to work alongside others who are asking the same ones — in schools, universities, and across the EdTech community. If this resonates, let’s find ways we can do that together.