Langcert Academic

The Problem with Language Certification Today

So here’s what we kept running into while building Marvel PTE. Students would ask us: “Can your AI score my LanguageCert practice test?” or “Does your platform work for general English certification, not just PTE?” And we’d have to say no. Not yet.

But the question stuck with us. Because the more we looked at the language certification space, the more we realised how far behind the technology is. Most certification exams still rely on manual scoring for writing and speaking sections. That means candidates wait days (sometimes weeks) for results. The tests themselves use fixed question banks with no adaptive difficulty, so a C1 candidate sits through the same beginner-level questions as an A2 candidate before things get challenging.

And the feedback? Generally a score. Maybe a band descriptor. Nothing actionable. Nothing that tells a learner what specifically to work on.

We’d already solved most of these problems for PTE practice with our AI scoring engine. The question was whether we could apply the same approach to certification itself. Actually, no. The real question was whether we should. Building an exam prep tool is one thing. Building the actual certification platform is a fundamentally different challenge.

We decided it was worth it.

What We’re Building

Langcert Academic is SIAGB’s second AI education technology product, built on the same engineering foundation that powers Marvel PTE (85,000+ users, 2 million+ scored tests). But where Marvel PTE prepares students for someone else’s exam, Langcert Academic is the exam.

The platform uses an adaptive assessment engine that adjusts question difficulty in real time based on how a candidate responds. If you’re answering B2-level questions correctly, the system moves you up. If you’re struggling, it recalibrates. The result is a shorter, more accurate assessment that zeroes in on your actual proficiency level faster than a traditional fixed-format test.

All four skills are covered: reading, writing, listening, and speaking. The AI writing evaluator assesses grammar, coherence, vocabulary range, and task achievement against CEFR standards. The speech analysis engine evaluates pronunciation, fluency, and spoken grammar. Everything is scored instantly. No human markers, no waiting period.

For institutes and educators, we’re building comprehensive dashboards: cohort management, progress tracking, analytics across student groups. The kind of tooling that lets a language school actually see whether their teaching methods are working, not just hope they are.

What We Learned from Marvel PTE

This practical foundation is what sets our approach apart. We’re not building Langcert Academic in a vacuum. We’ve spent years running a production AI assessment platform at scale, and that experience shapes every decision we’re making.

We learned that scoring accuracy isn’t just a technical problem. It’s a trust problem. If students don’t trust the AI’s score, they won’t use the platform. So we’ve invested heavily in calibration and transparency. We show candidates exactly why they received the score they did, with specific feedback on each skill area.

We learned that adaptive difficulty needs guardrails. Early versions of our adaptive system in Marvel PTE were sometimes too aggressive. A student would get a few questions right and suddenly face C2-level content that crushed their confidence. That balance between challenging and demoralising (it’s harder to get right than you’d think) is something we’ve refined over millions of scored sessions.

And we learned that institutes need different things than individual learners. An individual student wants to know their score. An institute wants to know whether their B1 cohort is actually progressing, which teaching methods correlate with improvement, and which students need intervention. That’s driven a lot of our dashboard design.

For Institutes and Educators

If you run a language school, a corporate training programme, or an academic institution, Langcert Academic is designed for you. We’re building it with institute-level needs from the ground up, not as an afterthought.

Early access partnerships are open now. If you’re interested in helping shape the platform (and getting first access when it launches), we’d love to hear from you. Just reach out through our contact page or explore our broader EdTech AI services.

We’re also working with organisations that want to build their own custom AI assessment tools on top of similar technology. If that sounds like something you’re exploring, let’s talk.