Back to Projects
EdTech, Language Assessment

Swoosh English

Built an AI system to grade and evaluate non-native English-speaking medical students during their spoken-English exams.

Swoosh English
UK

Key Results

93%

Grading accuracy

~50%

Reduction in assessment costs

~45%

Reduction in assessment time

The Challenge

Assessing spoken English for non-native medical students is expensive and time-consuming. Each exam needs expert human markers, which limits how many students can be assessed and how quickly results can be turned around — a real bottleneck for a growing cohort.

The Solution

We built an AI system to grade and evaluate students' spoken English during their exams, consistently and at scale.

Automated Spoken-English Grading

Evaluates non-native English-speaking medical students during their spoken-English exams.

Consistent, Scalable Assessment

Removes the cost and turnaround constraints of relying solely on human markers.

The Impact

Tangible Outcomes

Reached 93% grading accuracy against expert assessment

Cut assessment costs by ~50%

Reduced assessment time by ~45%

Enabled consistent grading at scale

Key Takeaway

Showed that a carefully built AI assessment system can match expert grading closely while dramatically lowering the cost and time of evaluating spoken language.

Need to Assess at Scale Without Losing Accuracy?

Let's discuss how AI assessment can cut cost and turnaround while keeping quality high.