Every candidate screened the same way,
scored, not just sorted

The process today

A high-volume hiring round usually means the same resume read many times over by whoever on the team has an hour free that day. One recruiter might weigh years of experience heavily, another might focus on specific tools listed on a resume, and a third might be more forgiving of a gap in employment history. None of this is written down as a rule, it is just how each person happens to read applications, which means two equally qualified candidates can get different outcomes depending purely on who opened their file first.

Volume makes the problem worse, not better. When a job posting brings in hundreds of applications, the honest version of what happens is that the first batch gets read carefully and the later batches get skimmed, simply because reviewer attention runs out before the applications do. A strong candidate who happens to apply on day four of a two-week window is not reading as carefully as one who applied on day one, even though nothing about the role or the criteria changed.

The result a hiring manager actually experiences is a shortlist that feels somewhat arbitrary, built from whichever applications got the most attention rather than a consistent read of all of them against the same bar, and no easy way to go back and check why a particular candidate was passed over.

What the agent does

The agent reads resumes and applications, extracts the fields that matter for the role, and scores each candidate against criteria your hiring managers define and weight, so a resume submitted on day one and one submitted on day fourteen get read the same way. For roles that benefit from it, the agent runs a chat-based screening conversation in Telegram or WhatsApp, asking the specific questions that would otherwise take a phone screen to cover.

Every scored candidate lands on a ranked shortlist with an approve or reject button for the recruiter, so the human decision is a quick confirmation against a score and reasoning that is already laid out, not a cold read of raw material. For candidates who are a clear non-fit against the criteria, the agent can draft a rejection email in your tone, which only goes out once a recruiter has approved that batch. Everything, scores, chat answers, application data and current status, lives in a candidate CRM so a hiring manager can see the whole pipeline rather than a pile of email attachments, and every recruiter can see a live log of what the agent did and why.

What stays with humans

The scoring rubric itself is written and owned by your hiring managers, not invented by the agent, and any candidate near the cut-off gets a human look rather than an automatic decision either way. Final hiring decisions, interview invitations for promising but unconventional candidates, and any rejection that is not a clear non-fit all go through a recruiter. The agent’s job is to make sure every application gets read the same way and surfaced consistently, not to decide who gets hired.

Guards

Every score comes with the reasoning attached, so a recruiter or a candidate who asks can see exactly which criteria drove a particular outcome, and the full history is kept in a live log any recruiter can audit. Rejection emails are held for approval rather than sent automatically, and the agent is built to rank and flag, not to make a final reject decision on a borderline application. The scoring rubric is tested against a sample of real past applications before launch, so the first live batch is not also the first test of whether the rules behave sensibly.

Price and timeline

Running cost is usually $20 to $70 a month in model usage depending on application volume, with a budget cap set before launch.

Related

Once a shortlist exists, interview scheduling picks up the booking, and employee onboarding takes over once an offer is accepted, so the same pipeline stays consistent from application to day one. For questions new hires ask once they are in, internal knowledge base Q&A answers from your own policy documents instead of repeating the same question to HR. The staffing agency recruitment bot case study describes a close relative of this build: a Telegram bot that interviews and scores candidates with a live panel for the agency’s admins. Broader context is on the AI agents service page and the automation-everything overview.

Want to see this against a real batch of your past applications? Get in touch and we will test the rubric before anything goes live.

Tired of doing this by hand? We can take the whole routine off your team, not just this step: Routine takeover, from $400 →