Case Study - AI - Education

On-Premise AI Admission Counselor

AI Admission Counselor

About the client

NED University of Engineering and Technology is a public engineering university. Each admission cycle, it handles questions from a large number of applicants about programs, eligibility, and campus facilities.

NED Logo

The Summary

NED University reached out to Providus to help its admissions team respond promptly and relevantly to applicant queries. We deployed a custom-built, on-premises version of DocAI on the university’s own servers and connected it to the document workflows of their admission teams so query responses are relevant and accurate. The AI assistant answers only from those documents, cites its sources, and asks a short followup question when an applicant leaves out a detail.

NED University student images
Premise AI Admission Counselor

The Challenge

Applicants had questions about eligibility, seats, merit, hostels, fees, and deadlines, but relevant information was scattered across different documents, and the admissions team had to answer every

  • Scattered admission information

    The answers were spread across many documents. Applicants had to search prospectuses and notices to find details on eligibility, seats, merit, hostels, fees, and deadlines.

  • Manual responses from staff

    The admissions team spent a large portion of their working day answering applicant queries manually. Most of these questions were repetitive, and the answers were available but scattered across notices, prospectuses, etc. This took up time their team could’ve utilized somewhere else.

  • Accuracy & data control

    The answers had to be accurate, and using managed AI platforms like Claude and Gemini was not an option as the documents had to stay within the university’s digital infrastructure.NED University wanted an AI solution to answer queries with context built on their internal documentation.

What We Delivered

We deployed a custom-built version of DocAI that’s being used as “NEDUET AI” by the university. The assistant runs on-premises with local models, Postgres, and Docker, ensuring both documents and intelligence stay within the university’s infrastructure. It is pinned to the admissions corpus, so it answers only from NED’s documents and cites them in every response. The assistant doesn’t assume or hallucinate. When an applicant leaves out a detail that changes the answer, the assistant asks a short clarifying question. 

  • PostgreSQL logo
  • LLM logo
  • DOC Ai
  • Docker logo

All Technologies Used

The Outcome

  • Faster ResponsesWe delivered NEDUET AI as a pilot, branded for the university. The tool became monumental for speeding up responses to applicant queries.

    Faster Responses Image
  • Answers With SourcesThe assistant answers only from the admissions documents and cites the source for each answer, ensuring accuracy.

    Answers With Sources Image
  • Clarifying QuestionsThe assistant asks a short follow-up question when an applicant’s query is missing a detail.

    Clarifying Questions Image
  • Data Kept On-PremiseAll documents and models stay on NED’s own infrastructure. This allows the university to ensure data privacy and compliance.

    Data Kept On-Premise Image

Featured Case Studies