Free guide · 2026 edition · 12-min read
The AI Recruiter Toolkit
How Indian recruiters cut resume screening from 4 hours to 30 minutes using AI prompts, scorecards, and a repeatable workflow. No fluff, no theory — only what works in 2026 hiring volumes.
01.Why this exists
Most Indian recruiters I have spoken to over the last 18 months are doing the same thing — opening Naukri or LinkedIn at 9am, spending 6 to 8 hours wading through resumes, and submitting a shortlist by 6pm that they themselves are not fully confident in.
The bottleneck is rarely judgement. It is volume. A senior IT role in Bangalore today gets 200 to 400 inbound resumes within 48 hours. A recruiter giving each one 60 seconds of attention is already at 6+ hours of work — and 60 seconds is not enough to evaluate a 7-year career.
This guide is the screening playbook I have seen the best teams quietly using. It does not require new tools to start — most of it is process work. AI is what makes the process scale.
02.7 AI prompts that cut screening 80%
These are the prompts I have personally tested with ChatGPT, Gemini, and Claude on real Indian-market JDs. Copy them directly. Replace anything in [square brackets] with your own values.
Prompt 1 — Generate the scorecard from a JD
Read this job description and extract the 5 most important "must-have" signals an ideal candidate would show on their resume. For each signal, write 1 line on what the resume should contain to score "strong" vs "weak". Job description: [paste JD here] Output as a markdown table with columns: Signal | Strong evidence | Weak evidence
Prompt 2 — Score a single resume against the scorecard
Score this resume against the scorecard below. For each signal, give a score 0-10 and one sentence of evidence from the resume. Scorecard: [paste output of Prompt 1] Resume: [paste resume text] Output: total score / 50, then table with Signal | Score | Evidence.
Prompt 3 — Compare two resumes side by side
Compare these two resumes for [role title]. Tell me which one is a stronger fit and why, in 3 sentences. Be specific about what tipped the decision — a project, a year of experience, a technology, a team size. Do not give vague opinions. Resume A: [paste] Resume B: [paste]
Prompt 4 — Identify red flags in 30 seconds
Read this resume and list every red flag a hiring manager would notice in a 30-second skim. Examples: gaps over 6 months without explanation, role tenure under 1 year repeated 3+ times, mismatched seniority labels, vague achievements without numbers. Resume: [paste] Output: bulleted list, max 5 items, no commentary.
Prompt 5 — Pull interview questions from the resume
Based on this resume and JD, generate 5 interview questions that would reveal whether the claims on the resume are true. Mix technical (2), behavioural (2), and situational (1). Each question should be answerable in 60-90 seconds. JD: [paste] Resume: [paste]
Prompt 6 — Detect AI-generated resume content
Read this resume and rate the likelihood that the achievements section was written by an AI tool. Score 0-100 and explain in 2 sentences. Look for: identical sentence structure across bullets, lack of specifics on team size or numbers, generic verbs like "leveraged", "streamlined", "spearheaded" used 3+ times. Resume: [paste]
Prompt 7 — Write the rejection email that does not feel cold
Write a 5-line rejection email to a candidate I screened for [role title]. The candidate had [strength], but did not have [missing requirement]. Tone: warm, specific, under 80 words. Mention something I noticed in their resume so they know we read it. Sign off as "Team [company]".
03.The JD scorecard template
Most recruiters write a JD with 15 nice-to-haves and then complain shortlists are weak. The fix is a scorecard with 5 must-haves and nothing else.
Here is the template. Fill it once per role. Reuse it for every resume that comes in for that role.
| Signal | Strong evidence | Weak evidence |
|---|---|---|
| 1. Core technical skill | Used in prod for 2+ years, with team size or scale numbers | Listed in skills section but no project evidence |
| 2. Years of relevant experience | Last 3 roles in matching domain or stack | Mostly tangential roles or recent pivot |
| 3. Outcome ownership | Numbers attached: shipped X, scaled to Y, reduced Z by N% | Vague verbs without numbers ("worked on", "involved in") |
| 4. Stability or progression | Either 2+ years per role OR clear progression in titles | 3+ jobs under 1 year in last 5 years without explanation |
| 5. Team or domain context | Specific team size, company stage, customer scale mentioned | No context — could be from any company of any size |
Why 5 signals and not 10: if you cannot articulate the 5 things that matter most, the JD itself is unclear and no recruiter (or AI) will produce a clean shortlist from it. The scorecard exercise also forces hiring managers to commit to what they actually want, instead of “yeah, all of those are important”.
04.Setting up your ATS shortlist filter
Whether you use Naukri, LinkedIn Recruiter, Hirist, or an in-house ATS, the rule is the same — let the system do the obvious filtering so you only see resumes that are at least keyword-relevant.
1.Set keyword Boolean search before sourcing
Use AND/OR/NOT operators to narrow the funnel. Example for a Senior React Developer in Bangalore:
("React.js" OR "React" OR "ReactJS") AND ("Node.js" OR "Express")
AND (Bangalore OR Bengaluru) AND (5+ OR 6+ OR 7+ years)
NOT (intern OR fresher OR junior)2.Lock the experience band
Indian platforms over-filter on years of experience. A “5-7 years” band misses 8-year candidates with the right skills. Use “5+ years” with an upper sanity check at 12, then let the scorecard rank within the band.
3.Filter on outcome words, not just titles
Senior Engineer, Tech Lead, SDE-3, Staff Engineer all overlap. Title filtering misses good candidates. Filter on outcome words: “shipped”, “led team of”, “owned”, “scaled to”, “production”. These appear in resumes of people who actually did the work.
4.Reject the obvious before opening the file
Set a hard reject filter for: same company > 3 jobs (suggests promotion stagnation), no role longer than 11 months in last 5 years (job hopper risk in BFSI/IT-services), or zero numbers in achievements (cannot quantify own work). These cost you nothing to filter out and save the recruiter from a bad call.
05.3 questions for cultural fit
Cultural fit calls usually take 30 minutes and produce nothing but “they seem nice”. These 3 questions, asked together in under 10 minutes, give you a real signal.
“Tell me about a time you disagreed with your manager. What did you do?”
Why it works: Reveals whether they will speak up or quietly resent. India-specific tip: if they say "I just did what manager said", that is a red flag for senior roles. You want someone who pushed back constructively.
“What is something your last team did really well that you want to bring with you?”
Why it works: Tests for self-awareness and active observation. Strong candidates have a specific example (a code review process, a stand-up format, a doc culture). Weak candidates say "good vibes" or "transparency".
“If we hired you and you had a free hand to fix one thing in your first 90 days, what would it be?”
Why it works: Tests for proactivity vs ticket-taking mindset. Strong candidates have a real answer with reasoning. Weak candidates say "I would observe first" — which is the safe answer they think you want, but tells you nothing.
06.Screening mistakes to avoid
Reading every resume start to finish
70% of resumes you receive are not a fit. Reading them fully wastes 6+ hours weekly. Skim the first half-page (current role + last role + top skill section). If those do not match the scorecard, reject.
Doing screening at 6pm
Decision quality drops sharply with fatigue. Block screening between 10am-1pm only. Anything that arrives after 1pm gets queued for the next morning. Yes, this means slower turnaround. The trade-off is shortlists hiring managers actually trust.
Trusting the resume more than the LinkedIn
Indian candidates often have polished resumes and stale LinkedIns (or the reverse). Cross-check every candidate you advance. If the LinkedIn job dates do not match the resume, that is a serious integrity flag.
Skipping the scorecard for "obvious" candidates
The strongest-looking candidate on paper is the one most likely to interview-bomb. Scorecard everyone the same way. The discipline saves you from one or two costly bad hires per year.
Letting hiring managers add requirements mid-pipeline
A “must also have GraphQL” added 2 weeks into sourcing means re-scoring 80 resumes. Lock the scorecard before sourcing starts. New requirements wait for the next role.
07.The repeatable screening workflow
Here is the workflow that takes your screening from “marathon every Friday” to “30 minutes a morning”. Adopt the steps in order, do not skip ahead.
- 1
Lock the scorecard (Day 1 of role)
5 must-haves, signed off by hiring manager, frozen for the role.
- 2
Set up Boolean ATS search (Day 1)
Save the search inside Naukri/LinkedIn so you re-run it daily without rebuilding. Take 30 minutes once. Saves 30 minutes every day.
- 3
AI pre-filter (Day 2 onwards, 15 min/day)
Run incoming resumes through Prompt 2 against the scorecard. Reject anything under 30/50. This kills 70% of the volume before you spend human time.
- 4
Recruiter reads top 30% (15 min/day)
Of what remains, the recruiter reads the resume properly — first page only — and assigns strong_yes / yes / maybe / no.
- 5
Hiring manager picks 5 from the maybes+yeses (5 min/day)
The recruiter sends the maybes and yeses with their scorecard evidence to the hiring manager. The hiring manager picks 5 to interview. No back-and-forth.
- 6
Interview, hire, retro
After every hire (good or bad), update the scorecard with what you learned. The next role for similar profile starts from a better template.
08.AI tools comparison
For step 3 of the workflow (AI pre-filter), you have three real options. Honest comparison:
| Tool | Best for | Trade-off | Cost |
|---|---|---|---|
| ChatGPT / Gemini (free) | <30 resumes/week | Manual copy-paste, no batch processing, no audit trail | ₹0 |
| ShortlistAI | 30-500 resumes/week, India-first | Newer product, fewer integrations than enterprise tools | ₹999-5,999/mo |
| Enterprise ATS (Workday, SuccessFactors) | 500+ resumes/week, large companies | Long onboarding, US-pricing, often shallow AI | ₹50k+/mo |
Honest recommendation: if you screen fewer than 30 resumes a week, ChatGPT/Gemini with the prompts in section 2 is enough. If you screen more than that, the manual prompt loop becomes a chore — that is the point where a tool like ShortlistAI (which automates prompts 1, 2, 4 in a workflow) saves real hours.
Try ShortlistAI free on your own JDs
5 free shortlists per month, no card. Paste a JD, upload resumes, get a ranked shortlist with reasoning in under 90 seconds.
Get started — no signup needed for demo09.Pre-flight checklist
Before you start sourcing for any role, confirm all of the following. If even one is missing, fix it before you read a single resume.
- Scorecard signed off by hiring manager (5 must-haves, no nice-to-haves)
- Boolean ATS search saved in Naukri/LinkedIn/Hirist
- Reject filters configured (years, location, hard skills)
- Daily screening window booked on calendar (10am-1pm only)
- Pre-filter prompt template saved (Prompt 1 + 2 from section 2)
- Hiring manager has 30 mins/day blocked for shortlist review
- Scorecard updated from last hire (what worked, what missed)
- Backup recruiter named in case primary is on leave
10.What to do next
You will get one email from me every 2-3 days for the next 2 weeks. Each one covers a specific case study or tactic that builds on this guide. If they are not useful, unsubscribe at the bottom of any email — no questions asked.
If you want to skip ahead and just see what AI screening looks like on your own JDs, the demo is here:
If you have questions or want to share what worked for your team, reply to any of the emails — they go to my personal inbox.
— The ShortlistAI Team