Most schools I visit have the same problem. Student data sits in half a dozen disconnected systems: one for attendance, another for assessment, a third for behaviour, something else for wellbeing. Each platform has its own dashboard, its own export format, and its own idea of what a "report" should look like. When leaders need a joined-up picture, they export to spreadsheets and start stitching manually.
This is the quiet infrastructure crisis in education. And as schools layer AI tools on top of these fragmented systems, the question of who owns your data, who can see it, and who profits from it becomes impossible to ignore.
The Problem with Proprietary AI in Schools
Commercial AI platforms offer convenience. They're polished, well-marketed, and quick to deploy. But they come with trade-offs that most schools haven't fully reckoned with.
- Vendor lock-in. Once your data lives inside a proprietary system, switching providers means starting from scratch. The platform owns the format, the integrations, and often the insights derived from your data.
- Opacity. When an AI tool flags a student as "at risk" or recommends a particular intervention, can you explain how it reached that conclusion? With closed-source systems, the answer is almost always no.
- Data sovereignty. Many commercial platforms process student data on servers you don't control, under terms of service that can change at any time. For international schools operating across multiple regulatory frameworks, this is a serious governance gap.
- Cost at scale. Per-pupil licensing models add up quickly, and the schools with the smallest budgets often get the least capable tools.
None of this means proprietary tools are inherently bad. Some are excellent. But schools deserve alternatives that put transparency and control first.
What Open-Source AI Infrastructure Looks Like
Open-source doesn't mean amateur. It means the code is publicly available, auditable, and modifiable. In education, a growing number of organisations are building serious AI infrastructure on open standards, and the results are already measurable.
The "Data Spine" Concept
The most compelling idea in open-source education AI right now is what some practitioners call a "data spine": a single, unified layer that automatically standardises information from all your existing EdTech providers and MIS platforms into one coherent view.
Instead of exporting attendance data from one system and cross-referencing it with assessment data from another, a data spine ingests everything automatically and presents it through real-time dashboards. Leaders get a single source of truth without the spreadsheet archaeology.
This isn't theoretical. Open Education AI, a UK non-profit backed by Purposeful Ventures, currently serves over 30 school groups across England, representing more than 600 schools and 400,000 students. Their platform connects to dozens of EdTech providers and all major MIS platforms, giving school leaders the kind of unified data view that used to require an enterprise-scale IT department.
Predictive Analytics Without the Black Box
Where it gets particularly interesting is predictive capability. Using open-source models, platforms like this can anticipate attendance risks and safeguarding vulnerabilities before they escalate, with accuracy rates above 93% in some implementations.
Because the models are open, schools can inspect exactly how predictions are made. When a system flags a student as being at risk of persistent absence, you can trace the reasoning back through the data. That matters enormously for safeguarding, for fairness, and for professional trust in the system.
Open-source AI doesn't just give you better tools. It gives you the ability to explain your tools, which is exactly what responsible governance demands.
Cross-School Collaboration
One of the most powerful advantages of open infrastructure is the network effect. When hundreds of schools contribute to the same data models (anonymised, of course), smaller institutions benefit from insights generated across a much larger dataset. A single-form entry school can access the same quality of predictive analytics as a large multi-academy trust.
This collaborative model was designed specifically so that any school, regardless of size, location, or technical capacity, can access the same calibre of AI-powered insight. For international schools operating without a national data infrastructure to lean on, this kind of shared resource is particularly valuable.
Why This Matters for International Schools
International schools face an amplified version of every challenge described above. They operate across different regulatory frameworks, often without centralised ministry data services, and frequently with smaller teams expected to cover more ground.
The open, standards-based approach addresses several pain points directly:
- Regulatory flexibility. Open-source tools can be deployed on infrastructure you control, in jurisdictions you choose. That simplifies GDPR compliance, data residency requirements, and the patchwork of national regulations that international school groups navigate.
- No proprietary lock-in. When your data infrastructure is built on open standards, you can switch components without rebuilding the entire system. That's not just a technical convenience. It's a governance safeguard.
- Algorithmic transparency. Research into AI bias in education, including work examining explainable AI for safeguarding, has uncovered significant algorithmic bias in predictive models that would have disproportionately affected the most vulnerable students. Open-source systems allow you to detect and correct these biases. Closed systems don't.
Governance First, Technology Second
The temptation with any new technology is to start with the tools and work backwards to the policy. With AI infrastructure, that's the wrong way round.
Before evaluating any platform, open-source or otherwise, school leaders should be asking:
- Where does our data live? Physically, legally, and contractually.
- Who can access it? Not just internally, but which third parties process your student data?
- Can we explain the outputs? If an AI system makes a recommendation about a student, can you articulate the reasoning to a parent, a governor, or an inspector?
- What happens when we leave? If you stop using a platform, do you retain full access to your data in a usable format?
- Is the model auditable? Can an independent party verify that the AI is fair, accurate, and free from harmful bias?
If a vendor can't answer these questions clearly, that's your answer.
Getting Started
You don't need to overhaul your entire technology stack overnight. A practical starting point:
- Audit your current data flows. Map every system that holds student data and document how (or whether) they connect to each other.
- Identify the gaps. Where are you making decisions based on incomplete or manually assembled data? Those are the highest-value opportunities for a unified data layer.
- Explore open-source options. Organisations like Open Education AI and Edequity AI offer infrastructure built on open standards, designed for schools rather than retrofitted from enterprise software. The UK's Open Education Analytics initiative is another resource worth investigating.
- Start with governance. Draft your data governance principles before you choose your tools. The technology should serve the policy, not the other way around.
The Bigger Picture
The decisions schools make about AI infrastructure now will shape the data environment students learn in for years to come. Choosing open, transparent, auditable systems isn't just a technical preference. It's an ethical position.
Commercial AI will continue to play a role in education, and some of it is genuinely good. But the default shouldn't be to hand over student data to the platform with the best sales team. Schools have a responsibility to ask harder questions about transparency, ownership, and fairness. Open-source infrastructure gives you the tools, and the standing, to ask those questions from a position of strength.
The question isn't whether your school should use AI. It's whether you can explain exactly what your AI is doing with your students' data.
Matthew Wemyss is an AIGP-certified AI in Education consultant and practising school leader. Book a discovery call to discuss ethical AI infrastructure for your school.