A recruiter at a mid-sized tech company learned a valuable lesson about AI automation this week: always double-check the numbers before hitting send.

The company's new AI-powered offer letter generator pulled salary data from the original job description ($85K) instead of the negotiated amount ($95K) that the candidate and hiring manager had agreed to. The recruiter, trusting the magic of automation, sent it off without reviewing the details.

The candidate, being a rational human who's not about to turn down an offer in this economy, accepted immediately. Signed, returned, and screenshot the offer letter for their records within 30 minutes.

When the Recruiter Realized the Mistake

The recruiter caught the error the next morning during their weekly pipeline review. Immediate panic. Slack messages to the hiring manager. Emergency call with HR.

"Can we just... withdraw the offer and send a corrected one?"

HR: "They already accepted. Legally, this is binding. We're either honoring this number or having a very awkward conversation about rescinding an accepted offer."

Legal got involved. Finance got involved. Everyone was involved except the AI system that caused this mess, because you can't put software in a performance improvement plan.

The Company's Options

Option A: Honor the $85K offer and eat the difference between what they promised the manager and what's on paper.

Option B: Go back to the candidate, explain the error, and hope they're understanding enough to accept the higher number without getting spooked about working for a company that can't get offer letters right.

Option C: Rescind the offer entirely and risk a lawsuit, terrible Glassdoor reviews, and LinkedIn posts about how this company treats candidates.

They chose Option B. The candidate, being both confused and amused, agreed to the corrected $95K offer. But not before asking "So should I be worried about payroll accuracy if we can't get the offer letter right?"

Fair question.

The Lesson Nobody Will Learn

The company promptly sent an email reminding recruiters to "carefully review all AI-generated documents before sending to candidates."

The recruiter who sent the original offer immediately had to generate three more offer letters that same afternoon. You know what they did? Trusted the AI again and barely glanced at the numbers.

Because nothing builds confidence like a system that's wrong 10% of the time but saves 90% of your effort.

The AI offer letter generator is still in use. It still occasionally pulls the wrong salary data. And recruiters are still playing Russian roulette with every offer they send.

Welcome to the future of recruiting, where we've automated our way into needing twice as much QA.