Helping educators grade with evidence

I have spent the past three years building curriculum, teaching, and assessing students in software and AI engineering. Throughout that time, we always found students who cheated and we caught them using plagiarism checkers, code similarity tools, and instructor intuition.

Then generative AI changed everything.

Suddenly, most submissions reflected AI-assisted work. Code was nearly perfect. Documents were polished. Presentations were well-structured. The tools we relied on could no longer distinguish who understood the material from who didn't. Worse, when every student produces quality work, instructors develop a false confidence in how much their students actually know.

The one approach that consistently worked was oral assessment, asking students to explain and defend their work in front of a review committee. When we focused on understanding rather than surface-level artifacts, it became clear who had done the work and who hadn't.

But when you are working with an institution that trains more than 10,000 students per quarter, you cannot do that manually for every submission.

That realization is what led to Pruuva. After weeks of discussion with colleagues, we arrived at one conclusion: the only sustainable answer is to move from assessing the artifact to assessing the student's understanding of it. Pruuva scales the approach that already works (oral checks, adaptive follow-ups, and structured evidence) so every student gets a fair opportunity to demonstrate what they know.

The tools that exist today (plagiarism checkers, AI detectors, proctoring software) were built for a different era. As AI models improve, those tools become less effective. Pruuva is designed to get stronger as the models get better, because it measures what no AI can fake: whether the student actually understands their own work.

My biggest motivation is personal. As an educator, I take pride in my students' achievements. But right now, I cannot be fully confident in what their credentials represent and that is a deeply uncomfortable position for any teacher. If we don't solve this, credentials will lose their meaning, graduates will struggle to prove their abilities, and educators will lose the trust that makes their work matter.