Most EdTech rebrands are cosmetic. New name, same curriculum, updated logo. What is happening at Empire AI is something different — and the distinction matters more than it might first appear.

There is a moment, somewhere in the lifecycle of a technology company, when the founders realise they are no longer building the thing they set out to build. The world has moved. The original premise — still perfectly defensible when they wrote it — has quietly become a constraint.

For Empire Code, a Singapore-based enrichment school that has spent years teaching children aged four to nineteen how to code, that moment arrived in earnest sometime in 2024. The school had built a solid reputation. Parents trusted it. Children were learning real skills. By most measures, it was working.

And yet. Across the technology industry, something structural was shifting in how software was actually made. The skills that Empire Code was teaching — Python, JavaScript, the mechanics of syntax — were the same skills that AI tools were beginning to perform cheaply, quickly, and at scale. The ladder that a generation of technology careers had been built on was, rung by rung, getting shorter.

“This is not a rebrand. Rebrands are what you do when the name is the problem. This was a rebuild — starting with the question of what it means to prepare a child for a world that is genuinely different from the one we designed this curriculum for.”

The result is Empire AI — same school, same students, fundamentally different philosophy. The name change is the least interesting part of the story.

The Problem with the Old Model

To understand what Empire AI is trying to do, it helps to understand what Empire Code — and most children’s coding education — was built around.

The dominant model for teaching children to code, which took shape roughly between 2012 and 2019, rested on a few intuitions that seemed reasonable at the time. Learning to code develops logical thinking. Syntax fluency is a transferable skill. The technology industry will keep rewarding people who can write software. Therefore, teach children to write software.

The model produced real results. Tools like Scratch and visual block programming genuinely help young children understand sequencing and conditional logic. Python is a well-designed language for learning. There is nothing pedagogically wrong with any of it.

The problem is not that the old model was wrong. The problem is that it was optimised for a job market that is now reorganising itself faster than most curricula can follow.

Data from SignalFire, a venture capital firm that tracks hiring across hundreds of millions of professionals, found that recent graduates made up just 7 percent of new hires at large technology companies in 2024 — roughly half the pre-pandemic figure. A 2025 survey by LeadDev found that a majority of engineering managers expected junior developer hiring to decline over the long term as AI tools allowed senior engineers to absorb more of the workload. The entry-level rungs of a technology career are not disappearing — but they are changing shape.

The tasks that used to define those entry-level roles — writing boilerplate code, building routine features, translating specifications into working syntax — are precisely the tasks AI performs most competently. What is harder to automate, and what employers are increasingly paying for, is the judgment required to direct those tools: knowing what to build, why it matters, whether the output is correct, and what to do when it is not.

That is a different skill than knowing how to write a for-loop. And it requires a different kind of education.

Why Now, Specifically

The timing of Empire AI’s transformation is not accidental. The past two years have produced a set of genuinely new facts about how technology is built.

In April 2025, Microsoft CEO Satya Nadella disclosed that between 20 and 30 percent of code in some of the company’s repositories was being written by AI. Around the same time, Google CEO Sundar Pichai told investors that AI was now generating more than 30 percent of the company’s new code — up from 25 percent just six months earlier. Meta’s Mark Zuckerberg projected that AI might handle roughly half of the development work on his company’s Llama AI models within a year. These figures come from earnings calls and public conferences, not press releases.

For the wider technology industry, the implications are still being worked out. For a children’s education school trying to prepare students for careers that are ten to fifteen years away, they are clarifying.

“The thing we kept asking ourselves,” one of the school’s team has said, “is whether a child who spends two years memorising Python syntax is going to be better off in 2035 than a child who spends those same two years learning to think clearly about problems, use AI tools fluently, and build things that actually work. We kept arriving at the same answer.”

The broader landscape provides context. Singapore’s Ministry of Education and IMDA have been running the Code for Fun programme for years, and from 2025 expanded it to include new AI for Fun modules covering generative AI, prompt engineering, and AI-integrated robotics. The Singapore government’s National AI Strategy 2.0 and Smart Nation 2.0 framework, launched in late 2024, commit substantial investment — including a S$120 million fund for AI adoption and a S$150 million Enterprise Compute Initiative announced in Budget 2025 — to building what Prime Minister Lawrence Wong has described as a nation that can harness AI not just use it. The direction of policy and industry are, unusually, pointing the same way at the same time.

What Is Actually Different

It is worth being specific here, because the EdTech sector has a long history of relabelling existing products. “AI-powered” appears in more marketing copy than it does in actual product decisions. The question worth asking about Empire AI’s transformation is not what they are calling it — but what has actually changed.

The answer is: quite a lot. And the changes are architectural, not cosmetic.

Robotics has expanded, not contracted

Physical computing — building things with sensors, motors, and microcontrollers that interact with the real world — has grown more central to the curriculum, not less. The reasoning is grounded in developmental research: cause-and-effect learning through physical objects builds a different kind of systems thinking than screen-based work alone. The robotics track now incorporates AI-enabled tools at appropriate ages, giving students experience with the intersection of hardware and machine learning that is increasingly relevant across manufacturing, logistics, and engineering roles in Southeast Asia.

Block programming has been reoriented

Visual coding tools for younger children remain — the research on their value for building computational thinking is solid — but the curriculum’s emphasis has shifted. Less time on the mechanics of getting code to execute. More time on the reasoning behind it: why this sequence and not another, what happens when the conditions change, how does this connect to the mathematics or physics the child is also learning. The aim is for a six-year-old’s programming session to build habits of thought that transfer, not just a familiarity with a particular interface.

Game development now includes business thinking

Students building games are also being asked to think about those games as products — who the player is, what makes something worth returning to, how a creative idea becomes something with real-world value. This is not a pivot to entrepreneurship education. It is an acknowledgment that the skills required to build something people actually use extend beyond the technical, and that children who develop them early are building a broader foundation.

App development has been rebuilt around how apps are actually made today

For older students, the track now incorporates working with AI tools as genuine development partners, connecting to live data sources, understanding APIs, and deploying projects to real environments where others can use them. The goal is not just to teach the theory of software development but to have students experience the full cycle — from idea to deployed, functioning product — using the tools that professional developers currently use.

“The question is not whether children should learn to code. The question is what coding education is actually for — and whether the way we have been doing it still serves that purpose.”

The most significant structural change is the retirement of syntax-heavy Python and JavaScript instruction as a core focus for younger learners. In their place: building using AI as a creative and technical collaborator. This means learning to describe a problem precisely, interpret what the AI produces, evaluate whether it is correct, push back when it is not, and refine the output into something that works. Understanding what AI is doing — the logic beneath the interface — is part of what is taught. What changes is the frame: technology as a way of thinking and creating, rather than a set of rules to memorise.

None of this is presented as making things easier. The school is explicit that working effectively with AI tools requires a deeper understanding of underlying systems than typing out syntax exercises ever demanded. It is a harder thing to do well. It is also, the school argues, a more useful one.

Why It Matters Beyond One School

Empire AI is one enrichment school in Singapore. It is not setting national policy, and it would be an overstatement to describe it as transforming an industry. What it represents, more modestly, is a particular kind of signal — and in EdTech, signals like this one tend to matter.

Singapore’s EdTech sector has grown significantly over the past decade. More than 790 EdTech startups have emerged since the Smart Nation initiative launched in 2014, and the country’s education technology market is projected to reach US$2.2 billion by 2027 on the back of strong government support and high digital adoption rates. The sector has also, in recent years, been wrestling with a question that Empire AI’s transformation makes unusually visible: what is children’s technology education actually for, and how should the answer change as the technology changes?

Across the Asia-Pacific region, governments and institutions are arriving at similar questions from different angles. In India, China, South Korea, and elsewhere, AI literacy is being integrated into school curricula at scale. The frameworks being developed now — what to teach, at what age, using what tools — will shape technology education for a generation. The enrichment sector, which moves faster than national curricula and has more freedom to experiment, has an unusual opportunity to demonstrate what works before systems-level decisions lock in.

More specifically, there is a question about the role of coding education that the sector has not fully answered. The argument for teaching children to code was always partly instrumental — it builds logical thinking, it prepares for technology careers — and partly about cultural access: understanding how the tools that shape your world are made. The instrumental argument has gotten more complicated as AI tools reshape what professional coding looks like. The cultural access argument, if anything, has gotten stronger. A child who understands how AI systems work, what they are good at, where they fail, and how to direct them is better positioned than one who does not — regardless of whether they pursue a technology career.

Empire AI’s approach is, in essence, a bet on the second argument. Teach the thinking. Use the tools. Build real things. Syntax will take care of itself, or become less important, or change again — but the capacity to understand and direct intelligent systems is, if the research is right, durable.

The Ground Reality Check

None of this should be taken as settled. Empire AI is in the early stages of a curriculum transformation, not the end of one. The proof points — whether children who learn this way develop the skills they are meant to develop, whether those skills transfer to the outcomes parents and students care about — will take years to accumulate.

The broader EdTech sector faces real headwinds too. Global EdTech venture funding contracted to roughly US$2.4 billion in 2024, the lowest level in a decade, as the post-pandemic investment surge corrected sharply. Enrichment schools in Singapore operate in a competitive and cost-sensitive market. Building a new curriculum from the ground up is expensive, slow, and uncertain in its outcomes.

There are also genuine pedagogical debates that the school’s approach does not resolve. Researchers and educators hold a range of views on how early AI exposure should begin, what safeguards are appropriate at different ages, and whether deprioritising syntax instruction creates gaps in foundational understanding that become apparent later. These are not trivial questions, and the field does not have consensus answers.

What Empire AI has done is make a considered bet in a specific direction, with a reasoned argument for why that direction is the right one given what the evidence currently shows. That is different from claiming to have solved the problem. The responsible version of this story is not “this school has figured it out” — it is “this school has asked the right question and committed to finding out.”

A Closing Observation

Somewhere in the history of technology education, the means became the end. Teaching children to code was always supposed to be about something larger — critical thinking, creative problem-solving, the ability to build the world rather than just use it. Somewhere along the way, for many providers, the curriculum became the point. Python in term one. JavaScript in term two. Repeat.

What is interesting about Empire AI’s transformation is not the specific curriculum choices, which will no doubt evolve. It is the decision to go back to first principles and ask what the curriculum is actually trying to do — and to be honest about the answer when the original design no longer serves it.

In an industry where rebranding is cheap and curriculum change is expensive, that is a harder thing to do than it looks.

Whether it works — whether the children who learn this way prove, in five or ten years, to be better prepared for a genuinely different world — is a question that cannot be answered yet. But it is the right question to be trying to answer. And the willingness to attempt it, rather than defend the existing model because it is the existing model, is something worth paying attention to.

For the broader EdTech sector, and for the parents making decisions about AI education for their children right now, the experiment is worth watching.

Empire AI is headquartered in Singapore and serves students across the region.

Register for a Free Trial Coding or Robotics Class

Empire Code offers trial coding and robotics classes for children aged 4 to 16, taught by MOE-registered and carefully selected educators who support early and primary learners with clarity, structure and genuine care.

A single session is often enough to see what your child naturally gravitates to – coding, robotics, or a little of both. Please complete the form below and our team will be in touch with you shortly to arrange a trial class.