AI Interviews for Campus Hiring in India: Can One Screen Thousands?

Does AI interviewing fit a campus drive?

Yes, for the first round, and only the first. A campus drive is the same question asked of thousands of near-identical candidates, which is the job a rubric-scored AI interview does well. It does not replace the technical round, the college relationship or the offer decision.

That is the short answer for a TA lead at an IT services firm or GCC. The rest of this page covers where a drive breaks at scale, what each step looks like with AgentR, what AgentR leaves alone and what a drive costs in rupees as of October 2026. AgentR is one such tool, so read its sections with that in mind.

Why does the first round of a campus drive break at scale?

Because volume is unpredictable and the first round is where fraud enters. A drive has to read a very large pool with the same yardstick, and the tests and interviews that do the reading are the ones candidates now game.

The numbers are large. Infosys reported 5.8 million job applications and 87,286 offers in FY26, according to company figures cited by People Matters (9 June 2026). The same report says Infosys postponed assessments for more than 20,000 applicants after detecting multiple instances of impersonation and assessment malpractice. Infosys CFO Jayesh Sanghrajka said in April 2026 that the company expects to hire at least 20,000 freshers in FY27 (Outlook Business, 24 April 2026).

Demand also swings. Wipro's CHRO Saurabh Govil told a press briefing that the company has no target for fresher hiring in the next fiscal and that it is "completely on demand" (Business Today, 17 April 2026). A drive you cannot size in advance is a poor fit for a tool priced per seat or per year.

How common is cheating in online tests?

Nobody has a clean number. A July 2026 NewsBytes item quotes HackerRank's Vivek Ravisankar as saying up to 35% of test sessions now get flagged for suspicious behaviour (NewsBytes). The item cites no original publication, so treat it as one executive's estimate of flagged sessions, and flagged is not the same as proven. HirePro's 2023 report ranks the common forms (someone sitting beside the test taker, impersonation, leaving the camera) but gives no rates.

What changes at each step of a campus drive?

Four things change: the criteria are written first, every application is read, every shortlisted fresher gets the same interview, and the recruiter sees evidence rather than a score alone. A person still decides at each point.

Step 1: the rubric

Write what a good first-round answer looks like before the drive opens. Structured interviews had the highest mean validity of the methods compared in the Sackett et al. 2022 meta-analysis (r = .42), ahead of job knowledge tests (.40) (SIOP summary). AgentR can draft criteria from the role, and your engineering managers should edit them.

Step 2: reading every application

AgentR reads every application in full against your criteria instead of the top of the pile. Its verification rates each claim as proven, described, mentioned or not found. For freshers the claims are thin by nature (projects, internships, a college), so expect many to land as "described". It is a consistency check that does not contact colleges or employers, and it is not background verification.

Step 3: one link, one interview

Candidates book through a single link, with no calendar coordination. AgentR runs a 20, 30 or 45 minute AI interview that walks through the candidate's own projects and scores the answers against your rubric. Proctoring is included, and its purpose is to keep the answers the candidate's own; this page does not describe how it works. If proxy candidates are your main worry, the AI cheating prevention page and our piece on AI interview cheating go further.

Step 4: the ranked shortlist

The recruiter gets a ranked shortlist with the evidence attached, down to each data point. Nobody is auto-rejected, and a recruiter can invite any candidate whatever the cutoff says. The same rubric applies to applicant one and applicant a thousand, which is the point of the volume screening approach.

What does AgentR not do for a campus drive?

It does not replace your technical round, run aptitude or coding tests, confirm a degree with a college, or manage your drive end to end. A person makes every decision.

AgentR is in private beta, has no customers to point to and has published no accuracy figures for campus hiring or any other segment. It is not an applicant tracking system: there is no pipeline view and no public API as of October 2026, so a drive that needs deep integration with your ATS or HRIS goes through a custom volume plan, which you arrange by talking to the team. Its verification does not replace a background verification vendor. If you visit three colleges and see 300 students, a rubric, a coding platform and your own interviewers will do, and a tool adds little. For a comparison of a chat window and a screening tool, see can a staffing agency just use ChatGPT to screen candidates; much of it applies here.

What does a campus drive cost?

At list rates as of October 2026 it is ₹25 per applicant reviewed and ₹10 per interview minute, with verification and proctoring included and GST extra (AgentR pricing). The monthly plan is ₹11,999 and converts to ₹12,000 of usage.

As an illustration, not a customer: a drive draws 5,000 applicants and 25% of them (1,250) clear your bar and take a 20 minute interview. Reading costs 5,000 × ₹25 = ₹1,25,000. Interviews cost 1,250 × 20 × ₹10 = ₹2,50,000. The usage is ₹3,75,000 at list rates, which is ₹75 per applicant across the pool and ₹200 per interview. A drive of this size is beyond one monthly plan, and the pricing page sends larger volumes to a custom plan with a volume discount, so use the figure as a list-rate reference and not as a quote. Change the assumptions and the total moves in proportion.

The comparison that matters is interviewer time. Without a screen, engineering managers sit first rounds for candidates a rubric could have ranked.

Is it worth it for your drive?

It depends on how many freshers a human has to see before the real conversation starts.

Under a few hundred candidates and your interviewers already read every profile. Write the rubric and skip the tool.

Thousands of applicants and managers lose weeks to first rounds. This is where a rubric-scored first round and a ranked shortlist with evidence earn their cost.

Your main fear is proxy candidates. A screen helps keep answers the candidate's own, but it does not replace an in-person final round, which some large employers have moved back to.

You are not sure. Take a drive you have already run. Put a sample of its applicants through the screen and see whether the freshers you hired, and the near misses, land where your managers would have put them.

Run one open campus role through AgentR. The first role is free, up to 50 applicants and 5 interviews.

See what the trial covers on the pricing page.