AI Recruiting Academy › Module 2: Finding candidates with AI
The short answer
AI can write a first line about each candidate’s real work in seconds. But Gmail and Outlook do not judge your email on how clever it is. They judge it on whether your domain is set up properly, how many people you email, how many addresses bounce and how many people mark you as spam.
So personalize from facts you can see on the candidate’s profile, send from a properly authenticated domain at human volumes, verify every address, and make it easy to opt out. Google asks all senders to keep their spam rate below 0.3%, and both Gmail and Outlook now enforce authentication rules for high volume senders.
Why does AI outreach land in spam?
AI makes it cheap to send a lot of email, and volume is exactly what inbox providers watch. Most recruiting emails that land in spam fail for reasons that have nothing to do with the words.
| Cause | Why it hurts | Fix |
|---|---|---|
| Domain not authenticated | Inbox providers cannot confirm the email really came from you | Set up SPF, DKIM and DMARC |
| Bounces from bad addresses | Signals a bought or stale list | Verify every address before you send |
| Spam complaints | The strongest negative signal there is | Target tightly and make opting out easy |
| Sudden high volume | A new or quiet domain that starts sending hundreds a day looks like a spammer | Start small and increase slowly |
| Identical messages | Hundreds of near copies look automated | Real personalization, not a merged first name |
| Heavy links, images and tracking | Common in marketing and phishing mail | Plain text, one link at most in a first email |
What do Gmail and Outlook require from senders?
Google’s email sender guidelines apply to everyone who emails a personal Gmail account. Microsoft published new Outlook requirements in 2025.
| Rule | Gmail | Outlook |
|---|---|---|
| Who it applies to | All senders, with extra rules above 5,000 messages a day | Domains sending more than 5,000 emails a day |
| Authentication | SPF or DKIM for all senders; SPF, DKIM and DMARC above 5,000 a day | SPF, DKIM and DMARC |
| Spam rate | Keep reported spam below 0.3% | Not a published threshold |
| Unsubscribe | One click unsubscribe for bulk marketing mail | A working unsubscribe link recommended |
| What happens if you fail | Rate limits, spam folder or rejection | Rejection since May 5, 2025 |
Most individual recruiters never reach 5,000 emails a day. The rules still matter, because the same signals decide where smaller senders land. You can check your own domain in a minute with our free SPF, DKIM and DMARC checker.
How do you personalize outreach with AI the right way?
There are three levels of personalization, and only the last one earns replies.
| Level | Example | Verdict |
|---|---|---|
| Merge fields | “Hi Sam, I saw you work at Acme.” | Everyone can tell it is a template |
| Flattery | “Your impressive background caught my eye.” | Says nothing, and AI makes it worse |
| Evidence | “You moved Acme’s payroll to Paylocity across six states last year, which is the project this role starts with.” | Specific, true and relevant to the job |
Use the method from Lesson 4 of Module 1. Give the AI the candidate’s profile, the role brief and your rules, and ask for one opening line that connects something real on the profile to the job. A prompt like this works well:
You are a recruiter writing a first email. Using only the profile and brief below, write one opening sentence that links one specific thing this person has done to the most important need in the role. Do not praise them. Do not mention anything that is not in the profile. If nothing fits, reply "no clear link".
The last instruction matters. It lets the model say there is no good hook instead of inventing one. Then keep the rule from Lesson 2 of Module 1: nothing goes to a candidate that you cannot see on their profile.
Done well, AI help pays off. LinkedIn’s Future of Recruiting 2025 report found that companies whose recruiters make the most use of its AI assisted messaging are 9% more likely to make a quality hire than those that use it least. That is a link, not proof of cause, but it points the right way.
How should you set up sending?
Authenticate your domain. SPF, DKIM and DMARC are DNS records your IT person or domain host can add in an afternoon.
Protect your main domain. Many recruiting teams send cold outreach from a separate domain that looks like their own, so a reputation problem cannot hit the domain their clients and employees rely on.
Warm up slowly. New domains and mailboxes should start with a small number of emails a day and grow over several weeks.
Verify first. Run every list through an email verifier, and recheck older addresses, as Lesson 3 explains.
Watch the numbers. Google Postmaster Tools shows your spam rate for Gmail. A sudden drop in opens or replies usually means you are landing in spam.
What does the law say about recruiting emails?
This is an overview, not legal advice. In the US, the CAN-SPAM Act covers commercial email, and the FTC’s compliance guide says it makes no exception for business to business email. An agency recruiter emailing hiring managers to win new business is sending commercial email. The main rules are simple: honest sender details and subject lines, your physical postal address, a clear way to opt out, and opt outs honored within 10 business days. Each violating email can cost up to $53,088.
A one to one email offering a candidate a job may not count as advertising under CAN-SPAM, but following the same rules costs nothing and protects your reputation. For people in the EU and UK, remember the duty from Lesson 3 to tell them where you got their details in your first message.
What should you check before you press send?
| Check | How |
|---|---|
| Domain is authenticated | SPF, DKIM and DMARC checker |
| Copy avoids spam triggers | Cold email spam checker |
| Every address is verified | An email verification tool |
| AI lines are true | Read ten at random against the profiles |
| Opting out is easy | A plain line such as “Reply no thanks and I will not contact you again” |
| Volume is sensible | No sudden jumps from a new or quiet mailbox |
Quick answers
Can Gmail tell that AI wrote my email?
Google’s published guidelines focus on authentication, spam complaints, formatting and volume, not on who wrote the text. What gets AI written email filtered is sending lots of near identical messages to people who did not expect them.
How many cold emails a day is safe?
There is no official number below the 5,000 a day threshold. Start low on any new mailbox, increase gradually, and let your bounce and spam rates tell you when to slow down.
Should I use LinkedIn InMail instead of email?
Use both. InMail reaches people whose emails you cannot find and does not touch your domain’s reputation. Email is cheaper at volume and reaches people who rarely check LinkedIn.
Do I need an unsubscribe link in a candidate email?
For a one to one message, a plain line inviting them to reply if they do not want to hear from you is usually enough. For bulk campaigns and business development email, include a proper opt out.
Explore in the directory
Module 2: Finding candidates with AI
- What is AI sourcing?
- Semantic search vs Boolean search
- What is contact data enrichment?
- How does AI rediscover candidates in your database?
- How do you personalize outreach with AI without landing in spam?
You have finished Module 2. Next up is Module 3, Screening and interviewing with AI.