Welcome to The Hero 🗞️. This is approximately a 2.5-minute read.

You ran the search. 

AI came back with 200 candidates…

Congrats! Your pipeline is packed.

Or is it…?

You enrich them. 

Spend credits. 

Send outreach. 

Book phone screens. 

Then you're sitting across from a "Senior Software Engineer" who hasn't written code since 2021 - but still carries the title.

The tool didn't fail.

It did exactly what it was designed to do: cast a wide net.

It failed because you treated the net like a filter.

AI sourcing tools are discovery engines.

They find candidates. They don’t always reliably verify them.

Skip that step - and every hour and dollar downstream gets wasted on people who were never right.

The full breakdown is just below - don’t miss it! 😉

Here are some of the best links I’ve found since last time I emailed you: 

 24 Best Applicant Tracking Systems: Full Comparison 2026 (link)

 Best Applicant Tracking Systems in 2026: Compared & Ranked (link)

Top 5 Staffing Trends to Watch for 2026 (link)

 9 Staffing Industry Trends to Watch in 2026 (link) 

10 Proven Employee Retention Strategies for 2026 and Beyond. (link)

2025 Retention Report: Employee Turnover Insights and Trends (link)

The State of Job Descriptions: Trends, Best Practices & AI-Powered Optimization (link)

10 Tips for Using Inclusive Language in Job Descriptions (link)

Every list an AI tool hands you is a set of claims. 

"This person matches." 

Maybe…

Or maybe the tool saw “React” in their headline and ignored that their last React work was three years ago.

Matched on title - missed that they left engineering for product management in 2023.

And bulk verification makes it worse… 🫨

Running 200 candidates through a "fit yes/no" check creates pattern-matching hallucinations.

The AI says yes because the candidate looks like other yeses - not because the evidence is actually there.

Here’s how you solve that…

Calibrate each batch.

Verify each candidate individually - read the full profile, confirm must-haves are supported by actual work history, not just listed skills.

If the batch is clean, scale from there.

If it's full of false positives, congrats - you just saved yourself from scaling garbage.

For every must-have, find the exact proof in their profile and logic to support it.

👉️ specific roles
👉️ specific projects
👉️ specific tenure

If the AI can’t point it out - the candidate isn't verified.

This one discipline saves more wasted outreach than any prompt trick ever will.

8/10 minimum to present.

Don't hand out 8s like participation trophies 🏆

The threshold has to hurt a little - or it means nothing. Be strict.

Same discipline. More volume - expand to your list of 200.

AI sourcing is the best discovery engine recruiting has ever had.

It's also the best false positive generator.

Build the verification layer - or keep wondering why your pipeline is full and your hires aren't.

Want to try this right now?

Copy / paste this prompt, and attach your JD:

"Run a high-precision candidate search for the JD below. Only count criteria that are explicitly evidenced in the candidate profile - if evidence is missing, ambiguous, or stale, treat it as unknown, not a pass. Don't trust title matches or keyword overlap alone. Export the full candidate JSON and verify against the JD, not the UI summary. Use this scoring: 10 = explicit evidence for every must-have; 9 = minor non-critical concerns; 8 = strong fit with modest limitations; 7 or below = don't present. Only show me candidates 8/10 or higher. Start with 15 candidates as a calibration batch."

See what comes back 😎

Then verify it yourself - before you expand the search or spend a single credit.

There are a few ways:

Or you can just reply to this email.

I reply to absolutely everyone who writes me back 🙂