An AI referral finder searches Google's public index of LinkedIn profiles, ranks employees at your target company by real overlap with you, and drafts an intro message you edit and send yourself. myjobb is India's first AI job agent and a universal auto-apply agent: it applies to jobs on Naukri, Foundit, Hirist and Instahyre, and on company career pages running Workday, Greenhouse, Ashby, Lever, Oracle Recruiting, SuccessFactors, iCIMS and other major ATS platforms. With myjobb, applying to a job is one click, and most applications happen automatically overnight, with the referral running in parallel.
Updated September 2026. This is the transparent, under-the-hood walkthrough of the finder, and a spoke of our job referral hub for India. By the myjobb Career Team, reviewed by a senior Indian tech recruiter.
Why is getting a referral so hard to do manually?
Because the hard part is not asking, it is finding the right person to ask, then writing a message they actually answer. Manual hunting means scrolling thousands of employee profiles on LinkedIn, guessing at shared colleges and past employers, and drafting each message from scratch. Most job seekers give up first.
Think about what a manual referral hunt at Razorpay or Swiggy looks like:
- You search the company on LinkedIn and get thousands of employee profiles.
- You scroll each one, hunting for a shared college, city, or past company.
- You cannot see full profiles beyond your 2nd degree without LinkedIn Premium.
- After an hour, you message the one person you found. Generic message. No reply.
Recruiters see the other side of this. Employees at Indian product companies get dozens of "please refer me" DMs a week. The generic ones get ignored. The ones that open with a real shared connection get answered.
So the problem splits in two. First, find the best person to ask for a referral, not just any employee. Second, give them a reason to reply within the first two lines. An AI referral finder handles both. And the prize is real: myjobb platform data shows referred candidates get hired at roughly 3-4x the rate of cold applicants, and referrals close in days, not months.
How does an AI referral finder actually find people?
It searches Google's public index of LinkedIn profiles for employees at your target company. myjobb never scrapes LinkedIn, never touches your LinkedIn login, and never asks you to upload a connections CSV. It queries publicly indexed profile data, then matches each profile against the master profile you built from your resume.
Here is the pipeline, using myjobb's referral finder as the example:
- You pick a target company, say PhonePe or Zoho.
- The AI queries the public web index for current employees there.
- It reads each public profile: education, past companies, city, role, team.
- It cross-references every field against your own background.
- It scores each person by how likely they are to reply and refer you.
No LinkedIn scraping means no account risk for you, and it matters for the person on the other end too. They are being found through information they chose to make public.
US-first tools work differently. JobRight has a solid Insider Connections feature that mines your own LinkedIn network for a job's hiring team. Refer.me runs a marketplace of volunteer referrers at Google, Meta, and Amazon. Both are built around US companies and US referral bonuses. Neither is tuned for the Naukri-to-LinkedIn workflow Indian job seekers actually run, and neither covers the Razorpay-to-Zoho company set.
How does AI rank the best person to ask for a referral?
It ranks people by predicted reply rate, using overlap signals between their profile and yours. An ex-colleague replies at 3.7x the rate of a cold stranger message, an alum at 3.4x, same surname plus city at 3.2x, and a hometown match at 2.1x, per myjobb platform data (2026).
- Ex-colleague. Who they are: Worked at one of your past companies · Reply likelihood vs cold DM: 3.7x
- Alumni. Who they are: Same college or university · Reply likelihood vs cold DM: 3.4x
- Same surname + city. Who they are: Likely community or family network overlap · Reply likelihood vs cold DM: 3.2x
- Hometown. Who they are: Same city or region you grew up in · Reply likelihood vs cold DM: 2.1x
- Recruiter / TA team. Who they are: Hiring is literally their job · Reply likelihood vs cold DM: High, replies expected
Two more factors adjust the ranking beyond raw overlap:
- Role and team relevance. A backend engineer referring you into a backend role carries more weight than someone in an unrelated function. The AI prefers people near the team you are targeting.
- Recency. Someone who joined recently remembers the interview loop. Someone who left your old company eight years before you joined is a weak ex-colleague match.
The output is not one name. myjobb surfaces 5-10 warmed referrers per target company, ranked and labelled with the exact overlap. You see why each person is on the list before you contact anyone. This is the "find best person to ask for referral" problem solved as a ranking problem: manual searching finds whoever LinkedIn shows first, AI finds whoever is most likely to say yes.
How does AI draft the referral message and the internal pitch?
It writes a short, story-led message in your voice, built from the real overlap it found. The draft names the shared connection, the exact role with job ID, and makes exactly one soft ask. It also writes a 40-60 word internal pitch for your referrer. You review both and send from your own account.
A good AI draft has three parts:
- Real overlap hook. What it does: Proves you did not mass-DM · Example line: "Saw you were at Freshworks 2021-23, I was on the CRM team till 2022."
- Exact role. What it does: Removes all work for them · Example line: "I am applying for the SDE-2 Payments role, job ID 4482."
- One soft ask. What it does: Easy yes, easy no · Example line: "If you feel comfortable after a look at my resume, would you refer me? No pressure either way."
The internal pitch is the step almost everyone skips. Referrers often stall because writing the "why this person" blurb for their company portal is work. The 40-60 word paste-ready pitch removes that last point of friction, so saying yes costs them two minutes.
Two honesty notes, because this is where AI tools oversell:
- AI drafts. You send. The message goes from your own LinkedIn or email account, after your edits. myjobb never messages anyone on your behalf. A bot-sounding template with your name on it still fails, so fix any line that does not sound like you. Full scripts live in our guide to asking for a referral on LinkedIn in India.
- AI cannot build the relationship. It finds the person and breaks the ice. Replying warmly, saying thank you, and updating them after the interview is on you. Our guide on asking for a referral without being awkward covers that human layer.
Why does an AI referral finder beat manual LinkedIn digging?
Because it compresses 2-3 hours of profile-by-profile searching into about a minute of setup, and it searches wider than your network. LinkedIn shows your connections; the public index covers every indexed employee. It also produces the two artifacts manual outreach skips: a personalised draft and the referrer's internal pitch.
- Find employees at target company. Manual on LinkedIn: 30-60 min of scrolling · With an AI referral finder: Under a minute
- Spot alumni and ex-colleague overlap. Manual on LinkedIn: Open each profile, hope · With an AI referral finder: Automatic, ranked by signal
- See beyond 2nd-degree connections. Manual on LinkedIn: Needs LinkedIn Premium · With an AI referral finder: Public index, no premium needed
- Write a personalised message. Manual on LinkedIn: 15-20 min each, quality varies · With an AI referral finder: Drafted in your voice, you edit
- Write the referrer's internal pitch. Manual on LinkedIn: Almost nobody does this · With an AI referral finder: 40-60 word blurb included
- Total per company. Manual on LinkedIn: 2-3 hours · With an AI referral finder: About 10 minutes including your edits
There is also a coverage gap you cannot fix manually. If you want to find referrals for jobs at 10 companies, manual digging does not scale on evenings after work. Public-index search does, because it is not limited to people you are already connected to, and it works for any company with employees in the index, not just firms on a marketplace's referrer list.
How do you run the finder end to end? A worked example
You pick a company, review a ranked list of 5-10 referrers with the exact overlap labelled, edit two drafts, and send them yourself from your own account. From naming the company to holding both drafts takes about a minute. Here is the full flow for a real target.
Say you are an SDE-2 in Pune targeting a backend role at Razorpay:
- Set the target. Open the referral finder and select Razorpay plus the job you want. Your master profile is already in myjobb from resume import.
- Review the ranked list. You get 5-10 referrers. Top of the list: an SDE at Razorpay who overlapped with you at your previous company for 14 months. Next: two COEP alumni on adjacent teams. Each name shows the exact signal.
- Read the drafted intro. A short story-led message naming the shared team, the exact Razorpay role, and one soft ask. Edit two lines so it sounds fully like you.
- Send from your own account. Copy it into LinkedIn or email and send.
- Hand over the internal pitch. When they agree, send the ready 40-60 word blurb. They paste it into the internal referral portal. Done.
- Follow up like a human. Thank them. Tell them when the interview lands. This relationship outlives this application.
Timing tip for India: send between 10 AM and 12 PM IST on weekdays. Referral DMs sent late night get buried under the morning pile.
Each big employer has its own portal quirks and hiring bar, so we keep company playbooks for getting a referral at Razorpay, at Swiggy, at PhonePe and at Flipkart. Freshers, your signals differ: college and hometown carry the weight when you have no ex-colleagues, and the playbook is in getting referrals as a fresher in India.
Can you pair the referral with an auto-applied application?
Yes, and this is the full loop. While the finder warms up an insider, myjobb's auto-apply agent submits the formal application: a fresh resume tailored to that JD in 10-20 seconds, then a real submission through your own logged-in sessions, mostly overnight. One click starts both.
The auto-apply side covers Naukri, Foundit, Hirist and Instahyre, plus company career pages running Workday, Greenhouse, Ashby, Lever, Oracle Recruiting, SuccessFactors, iCIMS and other major ATS platforms. Screening questions get reasoned answers in your voice, and a 9:00 AM IST report shows every application, question and answer. LinkedIn jobs appear in your ranked feed and you apply to them with one click yourself; LinkedIn auto-apply is on the roadmap, not live.
The pairing matters because referrals sometimes stall. People travel, forget, or change teams. With the application already in the ATS, a slow referrer delays nothing. With the referral landed, your submitted application jumps the queue. Either path can trigger the interview call.
On plans, listed on the pricing page: Free at ₹0 gets you the ranked feed and 10 auto-applies a month. Plus at ₹499/mo adds 100 auto-applies and referrals at 2 companies (10-20 referrers). Pro at ₹1,999/mo covers unlimited auto-applies and referrals at 10 companies (50-100 referrers). New to the category? Start with what an AI job agent is.
FAQ
How do I find the best person to ask for a referral?
Look for maximum overlap, not maximum seniority. Ex-colleagues and college alumni at the target company reply far more often than strangers, per myjobb platform data. Prefer people on or near the team you are targeting. An AI referral finder ranks publicly indexed employee profiles by these signals automatically.
Do referrals actually increase your chances of getting hired?
Yes, significantly. myjobb platform data shows referred candidates get hired at roughly 3-4x the rate of cold applicants, and referral processes tend to close in days rather than months. A referral moves your resume from the general pile to a human's desk, with an employee's credibility attached to it.
Is it OK to ask a stranger on LinkedIn for a referral?
Yes, if you personalise it and give them an easy out. Lead with a real shared connection like college or a past employer, name the exact role, and keep it under 100 words. Avoid "please refer me" with no context. Strangers with genuine overlap reply surprisingly often; pure cold blasts get ignored.
Does a referral guarantee an interview?
No. A referral gets your resume seen by a human faster and adds credibility, but you still need to clear screening. Pair the referral with a resume tailored to that exact JD. Companies also reject weak referrals quickly, so only pursue roles where you genuinely match the requirements.
Is my LinkedIn data safe with AI referral tools?
It depends on the tool's method. Some tools read your connected LinkedIn account data with consent. myjobb never connects to LinkedIn for referrals at all: it searches Google's public index of profiles, uploads no CSVs, and does no LinkedIn scraping. Every message sends from your own account, keeping you in control.
Ready to stop scrolling employee lists at midnight? Let myjobb's referral finder surface your 5-10 warmest referrers and draft both messages, while auto-apply lands the formal application overnight. You review, you send, you build the relationship.
