Your Next Bad Hire Is Being Made Right Now, Before You Post The Job
How founders can use AI to hire the right person, and which hiring decisions must stay human.
Listen to the audio here:
Most hiring mistakes are made before the job is posted. Founders hire for relief instead of a defined mission, so the smart use of AI in hiring is to write the brief and carry the coordination while a human makes the call.
Let me show you how that plays out, with one founder and four missions.
9 PM, A Full Inbox, And One Sentence
Dana runs a $2.5 million bookkeeping and advisory firm. (She’s fictional, but you’ve met her. You may have been her.) It’s 9pm. A client wants a status update, three onboarding packets are missing documents, a proposal is half-written, and two deadlines are wobbling.
So, Dana says what every founder says eventually. “I need to hire someone. An assistant.”
Spy agencies never recruit anyone because they’re tired.
They start with a requirement. The person comes second.
Read the sentence again. It has a feeling in it and it has a person in it. It doesn’t have a result in it.
That’s hiring for relief. And relief is a terrible job description. It produces a job ad that’s really a wish list (”manage my calendar, handle clients, coordinate projects, help with marketing, and basically think like me”), and it produces it fast, because Dana is tired.
What Does A Bad Hire Actually Cost?
SHRM’s 2025 benchmarking research put the average cost per hire for a non-executive role at $5,475, with screening and interviewing each averaging eight to nine days. That’s the cost of hiring when things go right.
When they don’t, the numbers get louder. A CareerBuilder survey run by Harris Poll found 74% of employers admitted to hiring the wrong person, at an average reported cost of $14,900 per bad hire. It’s a 2017 survey, so treat the dollars as a floor.
The more interesting finding is why. The top reasons employers gave were negative attitude (53%), not working well with others (50%), absences soon after hire (46%) and lacking the abilities they’d claimed (45%). Poor quality of work came in at 5%.
Most bad hires could do the job on paper. What failed was everything the résumé never tested.
Now put it in margin terms. Suppose one wrong hire costs you $15,000 you won’t get back. At a 30% contribution margin, you need $50,000 of new revenue just to cover it. (Illustrative numbers. Run your own.)
Why Do Founders Keep Hiring People Just Like Themselves?
Because without a defined mission, there’s only one picture of success in the room: the founder. We hire people who talk like us and interview the way we would. It feels like chemistry. It’s a mirror.
AI makes the mirror bigger if you let it. Reuters reported in 2018 that Amazon built an engine to rate résumés, trained on ten years of the company’s own applications, which came mostly from men. It learned to downgrade résumés containing the word “women’s.” Amazon scrapped it.
Feed an agent your past hires and it builds you a hall of mirrors.
The Four Missions
Here’s how I run hiring with clients. Four missions, in order, each handing its work to the next. Each has an AI agent, and each has a line the agent doesn’t cross.
Mission 1 · Brief (the Architect). The agent takes Dana’s messy workload and separates coordination from judgment. Then it lays out options before payroll: standardize the routine status updates, keep pricing and escalations with Dana, and explore a Client & Project Coordinator for what’s left. It drafts the mission brief: what the role owns, what it doesn’t, and objectives at 30, 60 and 90 days. When it doesn’t know how many hours something takes, it asks. Dana approves the need, the budget, and who will manage this person.
Mission 2 · Scout (the Scout). The agent turns the approved brief, and only the approved brief, into a job ad, a referral note, a short application with the same questions for everyone, an acknowledgment for every applicant and a visible pipeline. Dana approves the words, the pay and where it’s posted.
Mission 3 · Vet (the Analyst). The agent compares each finalist to the brief, flags what it doesn’t know instead of guessing, and drafts a work sample that looks like the real job: three client requests land at once and a project is late. What do you do first? In Dana’s case the candidate who felt like chemistry missed the late project entirely. The quiet one nailed it. Dana runs the interviews and makes the call.
Mission 4 · Handle (the Handler). In spy novels, nobody recruits an agent and walks away. The agent fills the offer into Dana’s approved template, builds the preboarding checklist and first week, and schedules check-ins against the same objectives written in Mission 1. Dana welcomes, coaches and tells the truth.
The agent carries the clipboard. A human makes the call.
What Should Never Be Handed To AI In Hiring?
The chair. The decision.
In 2023 the EEOC settled its first case involving AI hiring software. iTutorGroup’s software automatically rejected women 55 and older and men 60 and older, screening out more than 200 applicants. It surfaced when one applicant submitted two identical applications with different birth dates. Only the younger one got an interview. The settlement was $365,000.
So I hold three boundaries:
Answers, not faces. An agent can collect written or recorded answers to consistent questions. It shouldn’t judge anyone by their face, voice or body language, and candidates who need another format should get one.
Work samples, not invented personality tests. AI can draft a realistic exercise. A generated quiz is not a validated hiring instrument.
Consent, not snooping. Background checks run with the candidate’s permission through a proper provider, with the legal steps that come with it. No agent goes hunting through someone’s personal social media for reasons to say no.
How Much Time Does AI Save On A Hire?
In an illustrative model (one role, about 100 applications, 12 first-round screens, four interviews), a founder-led team spends roughly 37 hours by hand. With agents carrying the coordination, including the time to review their work, it drops to about 20. That’s 17 hours back per hire, a little over four hours a week across a four-week search.
The line that doesn’t move is interviews and the decision: six hours either way. Those were always the hours that mattered.
Filled Is Not The Same As Worked
SHRM found only 20% of organizations track quality of hire. Most count the seat as filled and move on.
Long-time readers know I call that a missing Verifier: the honest check on whether the thing you did actually worked. In hiring, it’s simple. At 90 days, hold the result up against the brief you wrote before you posted the job.
Spy agencies don’t recruit because they’re tired. Neither should you. No mission, no hire.
Where To Start: The Freedom Call
Before you post your next job, grab the Four-Mission Hiring Brief, a one-page checklist of what to define, what AI can handle and where your approval belongs.
And if you’re thinking about a hire right now, bring it to a Freedom Call. Bring one role, or the part of hiring that keeps getting stuck. We’ll look at the result your business actually needs, where an agent can carry the coordination inside Morgentiq™, and where you stay in the chair.
What to bring: one role you’re considering, or one hiring headache
What you leave with: a clearer mission, and a practical place to start
Book: callkoni.com
🎥 Watch the full episode on IconicTVLive.com
P.S. Next week’s mission: Your Secret Agent for Your Sales Team.
Same Tuesday, 2pm PST.








Discussion about this post