AI-Powered Application Evaluation Platform

EducationPrism9,200+ applications97% concordance

The stakes

When the Indian School of Business launched its iVi programme, it expected several thousand applications a year and gave the admissions team a clear mandate. Assess candidates consistently and objectively against a defined evaluation framework, return structured feedback to each of them, and stop placing senior faculty in front of applications they should never have had to review. Admissions decisions at an institution of this standing are examined, by the candidates who receive them and by the institution itself.

The gap

Evaluating at that volume by hand degrades in familiar ways. Reviewers differ in judgement, attention falls away over a long evaluation cycle, and differences in language fluency quietly distort an assessment that was meant to measure something else. All of it consumes the limited time of faculty who ought only to be seeing the final shortlist, and none of it produces feedback the candidate can use.

What we engineered and operated

We designed, developed and implemented PRISM, an AI-native application processing and document intelligence platform. It ingests and parses each submitted application and supporting information, extracting relevant content from documents of varying structure and quality and converting it into structured candidate information. The platform then conducts asynchronous, candidate-specific interviews as part of a structured AI-assisted assessment pipeline. Candidates are scored against configurable rubrics aligned directly with ISB's own evaluation criteria, and each receives a comprehensive report carrying quantitative scores, written feedback and insight. Faculty review an evidence-backed shortlist and conduct final interviews only on the candidates who reach it.

How it held up under scrutiny

The evaluation is defensible because ISB defines the rubric and the platform applies it identically to every candidate, and because the consequential judgement stays with people. Faculty conduct the final interviews, and PRISM's scoring was validated against theirs, reaching 97% concordance once faculty had completed their own assessment. We have remained accountable for the platform through post-implementation support across successive admissions cycles, with the rubric reconfigured as ISB's criteria evolve.

The outcomes we own
97%

concordance between PRISM’s scoring and the faculty’s own assessment

9,200+

applications assisted and evaluated

6,100

faculty-hours saved a year, across the admissions cycle

Tell us the number you need to move.

…outcomes that survive scrutiny.