Where AI actually helps in hiring
Hiring in a small Indian company usually looks like this: a post on Naukri, LinkedIn or Indeed, a few hundred applications, a founder or HR person skimming CVs late at night, and candidates lost because nobody called back in time. AI does not replace judgement here. It removes the slow, repetitive parts so the person hiring spends time on the 20 candidates who matter.
The four jobs where AI earns its place are: drafting clear job descriptions and screening questions, reading every CV against the same criteria, scheduling and reminding candidates, and keeping a clean pipeline so no candidate falls through the cracks.
What each AI step looks like
| Step | What AI does | What a person must still do |
|---|---|---|
| Job description | Drafts a JD and 5 to 8 screening questions from a short brief | Check salary band, location and must-haves are accurate |
| CV screening | Extracts experience, skills, location and notice period; scores against written criteria; writes a one-line reason | Review the shortlist and a sample of rejects every week |
| Candidate outreach | Sends WhatsApp or email messages with screening questions in English or Hindi | Approve templates; handle sensitive replies |
| Scheduling | Offers slots from the interviewer’s calendar, sends reminders, reschedules | Show up, and give feedback the same day |
| Interview notes | Transcribes and summarises interviews against a scorecard | Make the decision; never let AI decide hire or reject alone |
| Pipeline and CRM | Updates stages, flags stuck candidates, reports time-to-hire | Own the pipeline and close candidates out politely |
Working with Naukri, LinkedIn and job portals
Most Indian SMEs source from Naukri, LinkedIn, Indeed, Apna and referrals. These portals have their own applicant inboxes and paid recruiter products with search and filters. The practical setup is to export or forward applicants from each portal into one recruitment CRM or Google Sheet, where your AI screening and scheduling run in one place.
Avoid tools that scrape portals or automate actions inside your recruiter account. LinkedIn’s User Agreement bans bots and unauthorised automated methods, and portal accounts can be restricted. Use each portal’s official exports, email notifications, integrations or APIs where your plan offers them. Our guide to LinkedIn automation limits explains what is safe.
Do you need a recruitment CRM?
If you hire fewer than 5 people a month, a well-structured Google Sheet with stages, owner, next action date and source is enough, and AI can read and update it. Once you hire 10 or more a month across roles or locations, a recruitment CRM or applicant tracking system (ATS) pays off: it keeps every candidate’s history, avoids duplicate calls and gives time-to-hire numbers.
When you evaluate one, check that it imports from the portals you use, sends WhatsApp messages through an official WhatsApp Business API provider, stores data in your company’s account and lets you export everything if you leave. For broader CRM thinking, see CRM for small business in India.
Bias, fairness and the DPDP Act
AI screening copies whatever pattern you give it. If your criteria or past hires favour certain colleges, genders or cities, an AI tool will apply that bias faster. Write screening criteria about skills and experience the job truly needs, exclude names, photos, age, gender, religion and caste markers from what the model scores, and review a sample of AI rejects every week.
Candidate CVs are personal data under India’s Digital Personal Data Protection Act, 2023. The DPDP Rules were notified on 14 November 2025, with most obligations phasing in over 18 months. In practice: tell candidates how their data is used, use it only for hiring, keep it only as long as needed, secure it, and delete it on request. If you paste CVs into an AI tool, use a business plan that does not train on your data. Take legal advice for your specific case.
How to write screening criteria AI can apply fairly
The quality of AI screening depends almost entirely on the criteria you write. ‘Good communication skills’ is useless to a model and to a human. ‘Has handled inbound customer calls in Hindi and English for at least 1 year’ is testable. Split criteria into must-haves (reject if missing), nice-to-haves (add points) and red flags (send to a person, never auto-reject).
Ask the model to quote the line in the CV that supports each score. That makes every decision explainable to the hiring manager and to the candidate, and it exposes hallucinated matches quickly. When a strong candidate is rejected in your weekly sample review, fix the criteria, not the individual result.
| Criterion type | Example for a field sales role | What AI does with it |
|---|---|---|
| Must-have | Two-wheeler and licence; based in or willing to move to Pune | Rejects only if clearly absent, with the reason logged |
| Nice-to-have | B2B sales to retailers; Marathi speaker | Adds points and quotes the supporting line |
| Red flag for review | Gaps longer than 1 year; very short stints | Sends to a person; never auto-rejects |
| Never used | Name, photo, age, gender, religion, college prestige | Removed before scoring |
Tools Indian SMEs combine
- Sourcing: Naukri, LinkedIn, Indeed, Apna, employee referrals and your careers page.
- Pipeline: a recruitment CRM or ATS, Zoho Recruit, or a structured Google Sheet for low volumes.
- AI layer: ChatGPT, Claude or Gemini on a business plan, or a custom screening workflow built in n8n or Apps Script.
- Messaging: WhatsApp Business API through Interakt, AiSensy or Gupshup, plus email.
- Scheduling: Google Calendar or Microsoft 365 booking links with reminders.
A simple workflow for a 20-person company
- A hiring manager fills a 5-question brief; AI drafts the JD and screening questions.
- The job goes live on Naukri, LinkedIn and your careers page; applicants flow into one sheet or CRM.
- AI scores each CV against the written criteria and writes a one-line reason.
- Shortlisted candidates get a WhatsApp message with 3 screening questions and a slot picker.
- Interviews are booked into the interviewer’s calendar with reminders to both sides.
- After each interview, the interviewer adds a scorecard; AI summarises and moves the stage.
- Every Monday, HR gets a report: open roles, candidates per stage, stuck candidates, time-to-hire.
Checklist before you roll out AI in hiring
- Screening criteria are written, job-related and reviewed by the hiring manager.
- No protected characteristics are used in scoring.
- A person reviews every reject batch and makes every offer decision.
- Candidates are told how their data is used, with a contact to ask for deletion.
- AI tools are on business plans that do not train on your data.
- No scraping or unofficial automation of Naukri or LinkedIn accounts.
- Every candidate gets a reply, even if it is a polite no.
How Silvane can help
Silvane’s AI HR and admin team sets up CV screening, WhatsApp scheduling and a recruitment pipeline in your own Google, Zoho or CRM account, with human sign-off built in. Starter builds start from ₹50,000 with an optional run plan from ₹15,000 a month, tools at cost and GST extra. If you want candidate messages on WhatsApp, we connect them through WhatsApp automation on an official provider. The ₹199 discovery call (founding members, normally ₹15,000) maps which steps are worth automating for your hiring volume.
Sources: LinkedIn User Agreement · PIB: DPDP Rules, 2025 notified · LinkedIn Help: Prohibited software and extensions