Lecturers want to give good feedback. Students want to learn. Yet in most universities, both sides know the truth: there simply aren’t enough hours in the week. That tension, familiar to anyone who has ever marked a stack of essays, is exactly where Eduface steps in. It gives lecturers instant, detailed and domain-aware feedback on student writing, making high-quality assessment possible even when time is tight.
It began with a student’s frustration about a teacher not being able to provide the feedback they both wished for. Eduface has now grown into a fast-moving edtech startup with its own AI model, purpose-built for higher education. Today, Eduface is helping universities rethink how feedback, assessment and learning can actually work when time is no longer the bottleneck.
Born from a simple question: “Can you give me feedback?”
For CEO and co-founder Jeroen van Gessel, the idea emerged during his studies in Commercial Economics at Leiden University of Applied Sciences. When he asked a lecturer for more detailed feedback, he received a curt but honest response: “If I do this for you, I have to do it for 80 others. I simply don’t have the time.” Later, the lecturer admitted: “I wish I could give you this type of feedback — I just can’t.”
It was a problem Jeroen kept running into. He remembers that during his law courses in Leiden, the only real one-to-one conversation he ever had was with the receptionist when his student card stopped working — a clear sign, he says, of how little space there is for genuine dialogue in today’s lecture-heavy system.
A purpose-built AI model for education
With AI still relatively new at the time, he reached out to Menno Hahury, a technically minded AI student at Radboud University, and suggested building something that could actually close that gap. Menno joined full-time as CTO, completing his bachelor became a side project.
The team first experimented with off-the-shelf AI models, but soon discovered their limits: vague feedback, privacy risks, and models not trained for educational use. Menno explains: “General models are trained for millions of tasks. We needed a model specialised in feedback and grading. That meant building our own.”
Eduface now runs on an in-house model hosted on Dutch servers, trained specifically for academic feedback across six domains: from law and economics to STEM and health sciences. Unlike generic AI, it checks arguments, identifies factual errors, looks up jurisprudence, and comments directly in the student’s text. Eduface actively retrieves subject-specific information before reviewing a text, allowing it to fact-check claims and assess content with far greater precision.
Making high-quality feedback feasible
Eduface analyses both the assignment and the rubric, then generates highly specific, contextual feedback like academic writing issues, reference errors, and accuracy. The tool highlights passages in the document and suggests concrete improvements.
Jeroen illustrates the difference: “If you write a paper about parking an elephant in a living room, ChatGPT will talk about your introduction and conclusion. It won’t say: that’s impossible. Our model actually checks the content.”
Lecturers using Eduface often report that the tool returns more feedback than they can realistically provide themselves. Early pilots at Dutch universities show consistent improvements in writing quality, more formative feedback moments, and significant time savings.
Towards more meaningful education
Both founders see Eduface as a catalyst for a deeper shift in higher education. Jeroen argues that many exam formats exist not because they support good learning, but because they are practical to mark at scale. With AI easing the marking burden, institutions can move towards more active, practice-based learning.
For universities exploring programmatic assessment rather than one final exam, Eduface provides the missing infrastructure. “Educators have been wishing for this shift for years”, Jeroen says, “but it simply seemed too time-consuming. Tools like ours finally make grading based on multiple data points possible.”
He also hopes the tool frees lecturers to guide students in other ways: “I hope it allows lecturers to sit one-on-one with students more often, or in small groups. That’s where real learning and inspiration happens.”
Growing traction at home and abroad
Eduface is already used by several Dutch institutions, with strong results. In a recent survey with around 200 students, 89% preferred Eduface’s feedback over what they previously received. International interest is building fast too. Eduface recently won a UK tender, making them preferred supplier for universities across the country.
Through Briskr and BANN, Eduface is actively building its network. For the founders, the programme offers exactly what a growing edtech company needs: access to expertise and introductions to investors and institutions. They are currently exploring investment to expand their team and further expand their operations.
Digital sovereignty and the future of AI in education
As European institutions search for alternatives to US-based AI, Eduface positions itself as a sovereign, education-first solution. It trains its models without storing student data, and collaborates with lecturers and professors to refine domain expertise.
The long-term ambition is clear: personalised learning at scale. Eduface wants to help universities identify how each student learns best, where they excel, and where they need support. With institutions across Europe rethinking their approach to assessment, Eduface sees personalised learning as the next step in education.
