Building an AI teaching assistant grounded in real Further Education practice

I’ve spent most of my career working in Further Education. Teaching, learning technology, developing systems, and supporting lecturers as they try to make technology work in real classrooms rather than idealised ones.

That background heavily influenced how I approached building ClassCraft AI.

This wasn’t a project that started with the question “what can AI do in education?” It started with a more practical one. Where does time actually go for lecturers, and which parts of the job feel unnecessarily heavy?

In FE, a lot of effort sits around teaching rather than in teaching itself. Lesson planning, preparing resources, aligning content to units and outcomes, adapting materials for different cohorts, and documenting decisions all take time. That work is important, but it is also repetitive. Over the years I’ve seen experienced lecturers doing the same structural work again and again, often under time pressure.

ClassCraft AI was built to support that space.

The intention was never to automate teaching or replace professional judgement. Teaching is not a problem to be solved by software. What AI can do, though, is help with structure. It can generate starting points, outlines, and draft materials that lecturers can then adapt using their own expertise.

From a technical perspective, one of the main challenges has been avoiding generic output. General-purpose AI tools tend to produce content that looks plausible but lacks contextual grounding. In education, that becomes obvious very quickly. If a resource does not align with how a course is actually delivered, it adds work rather than removing it.

Because I work in the sector, the platform has been shaped by real workflows and feedback rather than abstract use cases. Features are added slowly and deliberately. If something does not save time or reduce friction for lecturers, it is not worth shipping.

Another important design decision was to focus on Further Education first, simply because that is the environment I understand best. That is not a statement about Scotland being unique, or about FE being fundamentally different elsewhere. It is about building something from a place of lived experience rather than assumption.

The longer-term principles behind ClassCraft AI are broadly applicable. Educators everywhere face similar pressures. But building something useful requires starting somewhere concrete and listening closely to the people who will actually use it.

The platform is live and under active development. It is being refined gradually, based on how lecturers interact with it in practice rather than how it looks in a demo. That pace is intentional. Education technology should earn trust through usefulness, not promises.

More information about the project is available at https://classcraft.ai.