Meaningful work is how future-ready skills grow.

We do not see project-based learning as an alternative to rigorous subject knowledge. Quite the opposite: before knowledge can be applied meaningfully, there has to be something worth applying.

Students should leave school knowing mathematics, science, history, literature, languages, technology, and much more. They should encounter ideas that are difficult, beautiful, useful, and sometimes far removed from anything they would have discovered by themselves. Teachers have an important role in deciding what deserves to be learned and in helping students understand things they are not yet capable of understanding alone.

The question for us is what happens after that knowledge is introduced.

Too often, the journey ends with proving that it can be remembered. We would like students to go further: to use what they know to make sense of situations that are less tidy than an exercise written to have one correct answer.

A mathematical idea becomes different when it helps you decide whether a proposal is viable. Scientific knowledge becomes different when it helps explain why something you designed did not work. Writing changes when there is an actual person whose mind you are trying to change.

The project does not replace the knowledge. It gives the knowledge somewhere to go.

And that matters because using an idea exposes understanding in ways that repeating it often does not. You discover quickly whether you really understand something when you have to decide when it applies, explain it to someone else, combine it with another idea, or recognize that it does not quite explain the situation in front of you.

Calling something a project does not make it meaningful.

A poster at the end of a unit can be a perfectly enjoyable activity, but the interesting question is whether the project itself asks students to learn. If all of the important thinking happened beforehand and the final task is mostly presentation, then the project is decorating the learning rather than carrying it forward.

We prefer projects in which the work itself keeps creating reasons to know more.

A student begins with an incomplete picture of the problem. New information complicates the first idea. A constraint makes the obvious solution impossible. Someone else sees the situation differently. An early attempt fails. The student has to return to something learned earlier, look for something new, or explain why one trade-off is worth accepting and another is not.

That is closer to the kind of uncertainty people encounter outside school. Real problems rarely arrive with all of the relevant information neatly supplied and a teacher standing nearby who already knows the answer.

That does not mean every school project has to involve a real company, a public exhibition, or a perfectly authentic simulation of adult life. A classroom can create meaningful stakes of its own. What matters more is whether the work contains enough context, ambiguity, constraint, and accountability that students cannot succeed simply by guessing what the teacher wants to hear.

We also care less about the format of the final result than about the thinking required to reach it. A prototype, presentation, report, campaign, model, debate, design, or video can all be worthwhile. None of them, however polished, tells us much on its own about what a student understood.

The work behind it matters.

Student agency is easy to celebrate in the abstract. In a classroom, it is more complicated.

Give students too little freedom and a project becomes another sequence of instructions to follow. Give them complete freedom and the work can become unfocused, unequal, or disconnected from what they actually need to learn.

We think the interesting territory lies between those extremes.

The educator remains responsible for the curriculum, the standard, the level of challenge, and the conditions under which students are working. Teachers decide what needs to be understood, where misconceptions matter, when a student needs support, and when someone is ready to be pushed further.

Within that structure, students can make choices that genuinely affect the work. They might decide which angle of a problem is worth pursuing, which evidence deserves more weight, how responsibility should be divided, what form a solution should take, or when an approach needs to be abandoned.

Those decisions are important because agency is not simply the feeling of being allowed to choose. It is the experience of having choices that carry consequences.

The same tension applies to personalization. We want students to receive different support when they need it, but we do not think personalization should automatically mean thirty students disappearing into thirty separate digital pathways. There is value in working on something shared: encountering other people's ideas, negotiating differences, helping someone else understand, and realizing that your own perspective is not the only one available.

Good structure and meaningful agency are not opposites. Often, the structure is what makes the agency possible.

See what this means for teachers →

We care about capabilities such as critical thinking, collaboration, communication, adaptability, creativity, curiosity, and judgment. But these words become vague very quickly if they are separated from the situations in which people actually need them.

Consider judgment.

We can tell a student that judgment matters. We can add it to a framework or a rubric. But judgment only becomes real when the student encounters a situation where the answer is not obvious: the information is incomplete, two priorities conflict, or several options are defensible and a choice still has to be made.

The same is true of many of the abilities we describe as future-ready skills. Collaboration becomes meaningful when people disagree and still have to move forward together. Adaptability matters when the original plan fails. Critical thinking becomes more than a label when a source seems convincing but does not quite stand up to scrutiny.

A good project can create those situations naturally because the student is trying to accomplish something rather than demonstrate a capability in isolation.

That does not mean participation guarantees development. A student can move through an entire project while avoiding the difficult thinking. One person can carry a group. A vague task can produce vague work. A polished outcome can conceal very little shared understanding.

So we are less interested in promising that projects “build” a particular skill than in designing experiences where students repeatedly have a reason to practise it, and where a thoughtful teacher can notice what is actually happening.

That distinction matters to us.

Imagine two groups giving presentations at the end of a project. Both are polished. Both look confident. Both appear, at first glance, to have done excellent work.

But the weeks leading up to those presentations may have looked completely different.

One group might have begun with a weak assumption, discovered evidence that challenged it, argued about what to do next, revised its approach, and eventually reached a conclusion everyone in the group can explain. Another might have divided up the slides, allowed one student to do most of the difficult thinking, and assembled an impressive-looking result without much shared understanding.

If we only see the final artifact, those differences are easy to miss.

That is why we are interested in the learning journey: not because every interaction needs to be captured, scored, or turned into data, but because decisions, explanations, revisions, feedback, reflection, and changes of mind can give an educator useful context about what happened along the way.

There is an obvious danger in taking this too far.

Not everything valuable about a student can be measured. Not every moment needs to become evidence. And more data does not automatically produce better judgment. Human development is too contextual and too complicated to be reduced to a perfectly precise dashboard.

We want technology to help educators see more where seeing more is useful, while remaining humble about what the evidence can actually tell us.

The goal is not to replace professional judgment with measurement. It is to give professional judgment richer context.

We believe deeply in meaningful, project-based learning. But belief is not evidence, and having a philosophy should not give us permission to stop questioning it.

In fact, one of the qualities we hope students develop is the ability to hold a view strongly enough to act on it while remaining willing to change that view when the evidence demands it. We should expect the same of ourselves.

This is particularly important in education, where broad labels often hide enormous differences in practice. Two schools can both say they use project-based learning while giving students very different experiences. An approach that works beautifully with one teacher, subject, age group, or community may work differently somewhere else.

So we are less interested in declaring that a particular method simply works than in understanding what makes an experience valuable, for whom, and under what conditions.

The same applies to Dwyl.

Our philosophy should guide what we build. It should never protect what we build from criticism.

If teachers tell us that something which made perfect sense on a whiteboard behaves differently in a real classroom, the classroom wins. If evidence challenges one of our assumptions, we should be prepared to change the assumption. And if a feature makes something more efficient while making the learning experience worse, efficiency is not enough.

Conviction should give us direction. Curiosity should keep us honest.

Generative AI makes a growing amount of intellectual work easier. That creates extraordinary possibilities for education, but also a difficult question: what happens when the work we are making easier is precisely the work a student is meant to learn how to do?

We do not think there is one simple answer.

If a student is trying to understand an argument, AI might help them question it from another perspective. If the task is to construct that argument themselves, having AI produce it removes something important. If a student is investigating a problem, AI can help surface information or suggest another direction. It should not quietly make every important decision along the way.

The distinction we keep coming back to is whether AI is supporting the learner's thinking or substituting for it.

The same principle applies to educators.

There is plenty of work around teaching that technology can help with: planning, organizing information, creating scaffolds, reviewing evidence, preparing resources, or making sense of what is happening across a complex project. We are excited about using AI to create more capacity for those things.

But teacher expertise is not an inefficiency to automate away.

Neither are the conversations in which a teacher realizes that a student understands more than their work initially suggested. The judgment that someone needs encouragement rather than another hint. The decision to let an unexpected discussion continue because the class has stumbled onto something worth exploring.

Education is full of moments whose value comes from context, relationships, and knowing the people in front of you.

Our shorthand for the boundary is:

AI assists. Students think. Educators decide.

We expect the details of that boundary to evolve as the technology does. The principle underneath it should be much more stable.

Read our responsible-AI principles →

The kind of education we are describing is not necessarily easier to provide.

Meaningful projects are difficult to design well. Different students need different kinds of help. Guiding a class through uncertainty can demand more professional judgment than delivering a tightly scripted lesson. Seeing the learning process rather than only the final submission introduces complexity rather than removing it.

That is precisely why we think technology has an interesting role to play.

We do not want to use AI to make education more efficient by stripping away the parts that make it human. We want to use it to make forms of teaching that are currently difficult to sustain more feasible.

We want students to know things deeply and learn what to do with what they know. We want them to encounter questions where the right answer is not immediately obvious, experience the frustration of an idea that does not work, learn to listen when someone sees the problem differently, and become increasingly capable of explaining not only what they think but why.

We want teachers to have the time, context, and tools to guide those experiences well.

That is the kind of learning we are building Dwyl around.

We are still early, and there are things in this philosophy that will become more nuanced as we learn from real classrooms. We see that as part of the work rather than a weakness to hide. The people closest to students should be able to challenge our assumptions while the product is still taking shape.

If you are thinking seriously about what learning should become in a world with AI, we would like you involved.

Join educator early access

Register school interest →