Coraltalk uses conversational AI to digitize large-scale university oral examinations

Generative AI has forced universities to rethink assessment methods. While current AI detection tools can flag text that may have been generated by a machine, they cannot verify if a student truly understands the material or can apply knowledge under new constraints. Coraltalk is addressing this gap by using conversational AI to bring individualized oral examinations into the digital classroom.

The platform functions as an automated substitute for one-on-one interviews. Instructors upload course materials and grading rubrics, and the software then uses the student’s submitted work as a basis for questioning. As students respond, the AI system probes their understanding by asking follow-up questions or introducing new variables to test critical thinking, a process that would otherwise consume dozens of hours for a single professor.

Co-founder Melissa Morgan stated that this technology makes deep assessment practical on a much larger scale. The platform allows instructors to review conversations in video format and transcripts, while importantly retaining the final authority over grading. However, technical challenges remain significant. Coraltalk CTO Adhikansh Mittal noted concerns regarding conversational latency, ensuring data privacy and security according to institutional requirements, and mitigating risks associated with AI hallucination or misinterpretation.

The startup has secured $768,000 CAD and gained traction in over 60 countries, including engagement with faculty at York University and James Madison University. This early adoption suggests a strong appetite among educators frustrated by the limitations of conventional assignments that fail to measure genuine learning.