The coming AI churn in EdTech: Why schools need to look beyond the badge

AI is now being added to almost every part of the EdTech market.

New vendors are entering the sector with AI-first products. Existing vendors are adding AI features into platforms schools already use. Some are doing this with care, structure and a clear understanding of the educational, safeguarding, privacy and operational environment in which schools work. Others are moving quickly because the market expects them to say they have AI.

This is where schools need to be careful.

The question is no longer simply whether an EdTech product uses AI. Increasingly, the question is whether the vendor has integrated AI in a planned, methodical and sustainable way.

Has the vendor understood the educational purpose? Has it considered age and stage? Has it assessed safeguarding risk? Has it considered data protection and children’s privacy? Has it tested for harmful outputs, bias, over-reliance, hallucination and inappropriate use? Has it given schools meaningful controls? Has it explained how AI is priced, monitored and governed? Has it thought about the long-term cost of delivering AI at scale?

If the answer to those questions is unclear, schools should pause.

We are predicting significant churn in the EdTech market as AI becomes more deeply embedded into products. Some AI-first vendors will grow quickly because they solve real problems. Some will struggle because they have built impressive demonstrations rather than sustainable products. Some established vendors will improve their platforms through thoughtful AI integration. Others will bolt AI onto existing products without fully understanding the technical cost, governance burden or school-level risk.

This is not unusual in technology markets. When a new capability appears, there is a period of excitement, experimentation and over-investment. New products appear. Existing products rebrand. Marketing moves faster than governance. Buyers are encouraged to believe that the presence of AI is, in itself, evidence of progress.

It is not.

AI is not a strategy. AI is not a safety framework. AI is not a procurement justification. AI is not a governance model. AI is a capability that needs to be applied to a clear problem, within clear boundaries, with clear oversight.

For schools, this matters because EdTech is not a normal consumer market. Schools are not buying technology for anonymous users. They are buying technology for children, teachers, operational staff and communities. The decisions they make affect learning, workload, privacy, safeguarding, digital safety, cyber security, budgets and trust.

The AI badge is therefore not enough.

AI will change the economics of EdTech

For many years, SaaS pricing has been relatively predictable. A school pays a licence fee, sometimes based on users, students, staff numbers, modules or school size. The vendor carries the cost of hosting, support and product development. The more schools that use the product, the more efficient the model can become.

AI changes that.

AI introduces a more variable cost model. Every prompt, response, summary, recommendation, generated lesson plan, assessment, translation, report, risk analysis or chatbot interaction may create a cost for the vendor. That cost may depend on the amount of information being processed, the complexity of the task, the model being used, the number of users, the length of the output and how many times the process is repeated.

This creates a different commercial reality.

A feature that looks inexpensive in a pilot may become expensive when used by hundreds of staff across hundreds of schools. A chatbot that works well in a demonstration may create significant cost when teachers begin uploading documents, asking follow-up questions, generating resources, retrying prompts and using it as part of daily work. A product that has been priced as a traditional SaaS licence may become difficult to sustain if the vendor has not controlled AI consumption.

This does not mean AI-enabled EdTech is a bad investment.

It means schools need to understand whether the vendor has designed the product to be sustainable.

The risk of bloated accounts

One of the issues schools should start to consider is how AI costs behave as more data is added to a product.

It is easy to assume that the cost increases because data is stored inside the LLM. That is not usually how these systems work. In most cases, the school’s data is stored in the vendor’s platform, database, document store or knowledge base. The AI model is then asked to process relevant parts of that data when a user asks a question or requests an output.

The cost risk therefore comes from use.

As more staff use the platform, more prompts are created. As more documents are uploaded, more content may need to be indexed, searched and retrieved. As more conversations take place, more context may be passed back into the model. As more workflows become AI-enabled, more AI actions take place in the background. The account becomes larger, richer and more useful, but it may also become more expensive to operate.

This is particularly important where vendors use broad context windows, long conversation history, repeated retrieval or agentic workflows where AI completes multiple steps before providing an answer.

A simple AI feature may involve one prompt and one response.

A more complex AI workflow may involve retrieval, classification, summarisation, drafting, checking, re-drafting and producing a final output. Each step may consume tokens. Each step may create cost. Each step may need to be controlled.

The danger is not simply that AI is expensive. The danger is that neither the school nor the vendor properly understands the cost curve until usage has already scaled.

Why some vendors will struggle

Some vendors will struggle because they are adding AI features without changing their product, pricing or governance model.

They may offer AI as an included feature, but without fair-use limits. They may give users open-ended prompts without understanding how usage will grow. They may rely on expensive models for simple tasks. They may retrieve too much data into every prompt. They may fail to separate low-risk drafting from high-risk decision support. They may not give schools admin controls, usage reporting or the ability to switch features on and off by role.

They may also underestimate the governance expectations schools will place on them.

Schools will increasingly ask whether AI is used in the product, what data is processed, whether children’s data is involved, whether data is used to train models, where data is hosted, how outputs are moderated, how harmful content is prevented, how bias is managed, whether staff can review outputs, whether students interact directly with AI, and what evidence the vendor can provide.

Vendors who cannot answer those questions clearly will lose trust.

This is why we expect churn.

Not because schools will reject AI. Schools will not reject AI where it is useful, safe, controlled and aligned to educational purpose. The churn will happen because schools will begin to distinguish between vendors who have integrated AI responsibly and vendors who have added AI quickly.

Schools need to ask better questions

Schools do not need to become AI engineers. But they do need to become more informed purchasers and governors of AI-enabled technology.

Before adopting or renewing an AI-enabled EdTech product, schools should ask:

What educational or operational problem does the AI feature solve?

Is the AI feature essential to the product, or is it an optional enhancement?

Who can use it?

Can students use it directly?

What data is processed by the AI feature?

Does it process personal data or children’s data?

Is school data used to train or improve models?

Can the school disable the feature?

Can the school control use by role, department or year group?

Does the vendor provide usage reporting?

How is harmful, inappropriate or inaccurate content managed?

What human review is required?

How does the vendor price AI usage?

What happens if usage exceeds expected levels?

Can the vendor explain how the feature is governed, tested and monitored?

These are not obstructive questions. They are reasonable questions.

They help the school understand whether the tool is safe, suitable, affordable and aligned to its strategy.

Better procurement, better governance, better budgets

AI will put more pressure on school budgets.

Some vendors will increase prices. Some will introduce AI add-ons. Some will move to usage-based pricing. Some will bundle AI into existing licences and then adjust pricing later. Some may withdraw features that are too expensive to run. Some may restrict usage. Some may change terms. Some may be acquired. Some may disappear.

Schools therefore need a clearer view of their EdTech environment.

They need to know which products use AI. They need to know which vendors are adding AI features. They need to know which contracts are due for renewal. They need to know where products duplicate one another. They need to know which tools are actually used by staff. They need to know which tools provide value and which simply create additional risk, cost or complexity.

This is where the Diamond Formation becomes important.

AI-enabled EdTech cannot be reviewed by one person in isolation. Academic teams need to understand whether the tool supports teaching, learning, assessment or workload in a meaningful way. Safeguarding teams need to understand whether it creates risks for children. Technology teams need to understand implementation, access, security and monitoring. Privacy teams need to understand data processing, vendor assurance and lawful use. Leadership needs to understand the direction of travel, the risks, the cost and the decisions being made.

The point is not to block innovation.

The point is to make better decisions.

AI must be intentional

The schools that manage this well will not be the schools that adopt the most AI tools. They will be the schools that adopt the right tools, for the right reasons, with the right controls.

They will know which AI use cases support their strategy. They will know which vendors have been assessed. They will know which staff are trained. They will know which tools are approved. They will know which risks remain open. They will know when leadership needs to be involved. They will know when a tool should be paused, rejected or reviewed.

They will also know when not to use AI.

That matters.

AI can support teachers. It can reduce administrative effort. It can help with drafting, planning, summarising, translating, analysing and organising information. It can help schools understand risk, evidence decisions and improve governance. But it can also create new risks if it is adopted because it is fashionable rather than because it is useful.

The best AI governance does not ask, “Does this product have AI?”

It asks, “Should this product have AI, and if so, under what conditions?”

The role of 9ine

This is why schools need an operating model for AI-enabled EdTech.

They need to be able to identify which vendors use AI. They need to assess privacy, safeguarding, cyber and AI risk. They need to connect vendor risk to application use, contract cost, staff guidance, training and governance actions. They need to evidence decisions. They need to report to leadership. They need to give staff confidence about what is approved, what is not approved and how tools should be used.

9ine supports this by helping schools govern technology, data and AI with confidence.

Through the 9ine platform, schools can bring together Vendor Management, Application, Contract, Privacy, Governance and Academy LMS. This gives schools a connected view of EdTech use, vendor risk, AI features, data protection, contract cost, staff training and leadership reporting.

The aim is not to make AI governance theoretical.

The aim is to make it operational.

A school should be able to see which applications use AI. It should be able to review vendor assurances. It should be able to link the tool to Records of Processing and DPIAs where required. It should be able to track contract renewal and cost. It should be able to publish approved guidance to staff. It should be able to assign training. It should be able to create tasks, risks and actions. It should be able to report to the Diamond Formation and to leadership.

This is how schools move from reaction to control.

Conclusion: the AI badge is not enough

AI will continue to reshape EdTech.

Some vendors will use it well. Some will not. Some products will become more valuable. Some will become more expensive without becoming more useful. Some vendors will thrive because they have planned carefully. Others will struggle because they have added AI without understanding the educational, technical, financial and governance implications.

Schools should expect churn.

They should also prepare for it.

The answer is not to avoid AI-enabled EdTech. The answer is to govern it properly. Schools need to understand what they are buying, what data is being processed, what risks are being created, what value is being delivered, what costs may increase and what controls are in place.

AI should support the educational mission of the school. It should not create unmanaged risk, hidden cost or unnecessary complexity.

The schools that succeed will be those that move forward intentionally.

Academic first. Safety always. Human in the loop. Governance with confidence.