The Risks of a Packaged AI Solution: the EDB Reminder and How to Choose

Last updated: 2026-07-22

With more than 790 schools each approved for HK$500,000 at the same time, the market's response was easy to predict: packaged solutions that "cover all five commitments in one contract" appeared everywhere, often priced close to the funding ceiling. In February 2026 the EDB reminded schools specifically to assess these with care. This article looks at the real risks of a package and at what makes a purchase safer.

What the EDB said

On 4 February 2026, in a clarification about the AI for Empowering Learning and Teaching Funding Programme, the EDB reminded schools plainly: when choosing a solution, assess with care whether it fits the school's own needs, and do not accept a market package wholesale. The background is that a great many packages had appeared, sold on the promise of "meeting the programme's requirements in one stop".

The point of the reminder is not that you cannot buy a package. It is that the responsibility to assess sits with the school. However a solution is packaged, it is the school itself that has to answer to the EDB for the commitments and for how the funding is used.

The four risks of a packaged solution

1. Uneven component quality

A package is priced so the strong parts carry the weak ones. One or two core products may be good, and the rest are there to make up the numbers. Compared item by item, the individual tools inside a package are often not the best in their class. Selling them as a bundle is exactly what takes away the school's chance to compare them one by one.

2. Lock-in to a single vendor

Sign a three-year contract and the school's bargaining power drops to zero. Find partway through that one component does not work and replacing it costs a fortune, because the vendor can simply say the package cannot be split. When the contract ends in 2028, exporting student data and teaching materials is another round of negotiation.

3. A poor fit with your school

A package is designed for the average school, but no school is the average school. Your three subjects, your teachers' starting point with technology and your students' language setting all differ from what the package assumes. The result of a poor fit: you buy ten tools, use two day to day, and the other eight become spending that is hard to explain in the report.

4. Thin evidence

Some packages even do the six teaching demonstrations for you: the vendor supplies ready-made lesson plans and the school uses them as is. The trouble is that the mid-term and final reports have to show what implementation actually looked like at the school. Ready-made lesson plans contain none of your students' work and none of your teachers' reflection, and they read as outsourced box-ticking. What the EDB wants to see is a change in the school, not the vendor's product catalogue.

A checklist for spotting a package trap

Is the quote close to the HK$500,000 ceiling without asking about your school? (A sensible solution starts from need, and the price varies from school to school.)
Can you buy the parts separately? Are the individual prices reasonable once split?
Can each component be trialled on its own, or can you only sign for the whole bundle?
How much of the done-for-you part, such as lesson plans, activities and reports, is actually school-specific?
What are the terms for pulling out partway through or swapping a component?
How are student data and teaching materials dealt with when the contract ends?

An evidence-led alternative

Break the purchase into three steps. First, work back from the five commitments to a list of needs. Then compare at least three options for each need, scoring them on four criteria: fit with the curriculum, Cantonese support, data policy and total cost. Finally, trial the winner on a small scale and use real usage evidence to decide whether to expand. This takes a few weeks longer than signing one package contract, but in return every item of spending has comparison evidence behind it. When the report needs to say public money was used with care, that evidence is a ready-made footnote.

If a school does not have the staff to compare the market itself, it can bring in a consultant who is independent of any package. The comparison should start from your school's needs and the conclusion should be backed by evidence, rather than starting from one particular product catalogue.

Common questions

Is a packaged solution always a bad thing?

Not necessarily. If every component in a package stands up to being compared one by one, can be bought separately and can be trialled first, then it is simply good tools that happen to be bundled. The risk is not the package as a format. It is the decision to buy the whole thing without comparing. That second thing is what the EDB was warning about.

We have already signed a package contract. What now?

Take stock first: which components in the package are actually being used? Put your resources into using those few well and producing evidence from them, and talk to the vendor about swapping the idle components or adding training. The report cares about real results on the ground, and focusing in time is wiser than trying to justify every item of spending.

Criteva is a vendor too. Can this article be trusted?

Criteva's position is written into the method: we argue that any tool, ours included, should pass the same set of comparison criteria and prove itself with evidence from a small trial. The checklist in this article treats every vendor the same. Schools are welcome to use it on us.

Sources

Want to know your school's next step?

Tell us where your school is with the funding programme. We help with evidence, from choosing tools to filing the reports.

Book a free school consultation Request a proposal