EdTech for Learners, Not Just Users: The Era of Evidence-Based Infrastructure
Returning to the BETT Show in London this year prompted a moment of longitudinal reflection. When I first walked the floor of the Olympia Hall back in 2009, the industry was still in a phase of exploration. We were asking, “what can this technology do?” The focus was on siloed technologies, novelty, and the sheer possibilities of digitization. Seventeen years later, the question hasn't changed so much as the standard for answering it. “What can it do?” now carries a more incisive subtext of “does it actually help?”
After a week of dialogues at BETT with policymakers, researchers, and global partners, and drawing on what my colleague Nicole observed at UNESCO's Global Education Coalition meeting in Paris a month later, it is clear that the Wild West era is over. The two gatherings couldn’t have been more different in audience or register, but the underlying diagnosis was the same. We’ve entered a new phase of market maturity defined by three converging trends: the shift from generative potential to pedagogical value, the rise of digital intentionality in policy, and the establishment of evidence as a requirement for scale.
Operationalizing AI for education
While artificial intelligence remains the dominant topic of conversation, the discourse has matured. We’ve moved past the initial excitement of rapid prototyping built on surface-level functionality, toward a demand for learner-ready infrastructure. In discussions with major cloud providers and assessment leaders, a consensus emerged. The winners in this next phase will not be those who simply wrap large language models (LLMs) around generic interfaces. Success will be defined by those who can demonstrate impact across three connected layers: verified academic benefit for students, measurable efficiency gains for the institutions serving them, and the sustainability that lets both continue past the pilot phase. The goal is no longer just to build tools that are capable of answering a question, but to build infrastructure that is rigorously engineered to support the learner.
Digital intentionality
Perhaps the most striking observation from London was how closely global strategic conversations mirrored the US policy landscape. In the U.S., we are seeing a legislative push toward learning and employment records (LERs), which move the economy away from static credentials and toward verified, granular skills. Simultaneously, there is intense scrutiny regarding screen time and device usage in schools. This aligns with the sentiment I heard from several European education ministries. One Nordic education minister described the concept of digital intentionality—the deliberate practice of choosing when a digital tool genuinely serves a learning moment and when it does not. The stance isn’t anti-technology, it’s anti-default. The message from policymakers is clear—edtech innovation includes the discipline of knowing when not to use technology.
This brings us to a distinction the industry can no longer afford to blur. Consumer tech is optimized for engagement and passive consumption. Educational tech, on the other hand, has to be designed from the start with pedagogical guardrails, privacy standards, and interoperability in mind. The difference matters, but only if the people making procurement decisions can actually see it. That’s the tougher problem.
The evidence exists in some contexts. What’s missing in many systems is the capacity to read it, trust it, and understand it enough to act on it. Think of it as evidence literacy—it’s truly the infrastructure everything else depends on when making informed decisions about edtech usage. A tightening regulatory environment around privacy, safety, and efficacy only works if the people enforcing it can tell a credible claim from a polished one.
Evidence as infrastructure
In a mature landscape, trust is the primary currency. But as the pace of AI innovation has outstripped traditional evaluation frameworks, we face a methodological friction point unlike any other in the history of education.
Traditionally, the field has leaned on randomized controlled trials as the gold standard, and for good reason. A well-designed RCT remains the most rigorous way to answer a stable question about a stable intervention. The issue is that AI features rarely sit still long enough for a two-year study to catch them. By the time a traditional trial reports out, the functionality it evaluated may be three or four versions obsolete. The finding may be rigorous, but it’s answering a question about a product that effectively no longer exists.
The pivot toward testbeds, including the UK government's recent investment, is not a rejection of rigor. It’s an attempt to match the cadence of evaluation to the cadence of the thing being evaluated. Evidence has to become a continuous R&D process rather than a static marketing asset or a downloadable PDF. For evaluators, the question morphs from “did this work?” to “is the study I’m reading still about the same tool?”.
The path forward
The education industry is pivoting from access to impact, and this shift demands a new level of nuance. We’re moving past the era of generic claims where a tool is simply labeled effective—the context is what matters. We can’t just ask, “Does this technology work?” The question becomes, “Does it work here, for these learners and educators, under these specific conditions?” We must move beyond building tools for educators to building them with educators. It should be a two-way partnership where real-life learning spaces shape the technology and not the other way around.
Whether it’s navigating the digital intentionality of a screen-free policy or adapting to fast-changing AI models, the learning tools that define the next decade won’t be the ones that claim to solve everything for everyone. They’ll be the tools that understand their specific place in the learning journey, and can clearly prove their value within the complex, messy, and beautiful context of the classroom. Which means ultimately, the people deciding whether or not to trust them can be confident they’re making the right choices for learners.
