We Used AI to Run a Full Hiring Loop. Here’s What It Did.. and What It Couldn’t.

I want to tell you about a hiring process I ran recently with client that changed how I think about where AI belongs in decisions that matter.

We were hiring for a first-of-its-kind role at a client.. a Strategic Account Manager, the first dedicated AM they’d ever brought on. Five candidates. Four rounds of interviews. Six internal stakeholders. A specialized industry where relationship depth matters as much as the resume.

I used Claude as an end-to-end intelligence partner across the entire loop.

Not to make the decision. To make sure the humans making the decision had everything they needed.. and that nothing important got lost between rounds.

The Problem With Most Hiring Processes

Here’s the failure mode I’ve watched play out at every company I’ve worked with.

Round 1 surfaces something interesting about a candidate. A tension, a gap, a signal worth tracking. The interviewer makes a mental note. By Round 3, that note is gone.. overwritten by everything else that happened in the two weeks between. The Round 3 interviewer asks what feels natural, which often re-covers ground that Round 1 already mapped. The CEO does the final interview with a verbal summary from the hiring manager and whatever impression they form in 45 minutes.

By the time a decision gets made, nobody has a complete picture. They have fragments, impressions, and whatever the most recent interviewer said most confidently.

That’s how companies make expensive hiring mistakes.. not because the right signal wasn’t there, but because nobody had a system to carry it forward.

What AI Did in This Process

Before every interview, a candidate prep kit was generated from the resume and job description. Not a generic list of questions.. a SWOT analysis specific to that candidate, an experience gap analysis tied to the role, and a set of questions designed to probe the gaps identified before the call even happened.

For one finalist.. I’ll call him Candidate A.. the pre-call analysis flagged that his background was strong on the leadership and coaching side but unproven at the individual contributor execution level the role required. That specific tension became the thread we tracked across all four rounds. It shaped what Round 1 asked, what the handoff brief told Round 2 interviewers to probe, and what the project presentation was designed to surface.

After each round, the transcript was analyzed and a structured handoff brief went to the next interviewer.. what was covered, what the signal said, what remained open. Not a summary of strengths and weaknesses. A brief that explicitly said: here’s what we already know, here’s what we don’t, here’s what you should go find out.

Before the CEO round, a written briefing was drafted that gave three-round context on both finalists and specific guidance on what angles the prior rounds hadn’t had time to cover. The CEO walked in prepared to go deeper, not to start over.

What AI Could Not Do

Here’s the part I want to be equally honest about.

The framework surfaced signal. It could not decide what that signal meant.

When one candidate had a competing offer with a tight timeline, that required human judgment on timing, offer construction, and what the company could credibly put forward. AI kept the risk visible. It couldn’t resolve it.

When the peer interviewers spent time with a candidate, their gut read on whether they’d actually want to work with that person.. whether the collaboration would feel like partnership or friction.. that didn’t come from a transcript. It came from sitting in the room.

And the hire itself.. the decision to extend an offer to Candidate A over a strong alternative.. that was a human call, made by a human, based on four rounds of evidence that the AI layer helped preserve and organize. Not a recommendation the AI made.

That line matters. AI as intelligence partner is a completely different thing from AI as decision-maker. The moment you blur that line in hiring, you’ve created accountability gaps that will eventually cost you.

The Signal That Would Have Been Lost

Here’s the thing I keep coming back to.

Candidate A’s resume looked like a coaching and team-building background. On paper, he looked like someone who’d grown out of individual execution. That framing would have killed his candidacy in a traditional process where the hiring manager does a phone screen and forms an impression before a structured loop begins.

The transcript analysis from Round 1 told a different story. Specific examples emerged.. how he’d managed a 20-year client relationship through a near-cancellation, how he’d built a structured success plan without a playbook, how he’d proactively created executive relationships that weren’t in his job description. None of that was on the resume.

The AI layer didn’t discover those things. The interviewer did. What the AI layer did was document them with enough structure that they carried forward.. accurately and completely.. to every subsequent decision-maker.

In a traditional process, that insight lives in one person’s head and fades. Here it became part of the evaluation record that everyone in the loop was working from.

The Outcome That Surprised Me Most

We hired Candidate A. But the outcome I didn’t expect was what happened with the runner-up.

In a traditional hiring process, candidates who don’t get the role are effectively gone. The evaluation exists in someone’s memory, and when a future role opens up, you’re starting the process over.

Because this process was documented, the runner-up came out of it with a full evaluation record, a clear profile, and a pre-warm standing for the next role that opens. That’s zero additional sourcing cost for a future hire.. on a candidate who was already through a rigorous four-round loop.

That’s the compounding value of structured hiring intelligence. It doesn’t just improve the hire you’re making. It builds the pipeline for the hires you haven’t started yet.

What This Tells Me About AI in High-Stakes Decisions

The instinct when you hear “AI-assisted hiring” is to think about bias, automation, and replacing human judgment. I understand that instinct. It’s worth taking seriously.

But that’s not what this was.

This was AI doing what AI is genuinely good at.. synthesizing information across a long, complex process and making sure nothing important gets lost. And humans doing what humans are good at.. reading a room, making judgment calls, deciding who they want to bet on.

The question isn’t whether AI belongs in decisions that matter. It’s whether you’re using it for the right things and keeping humans accountable for the things AI can’t own.

In this case, we got that balance right. The hire is in seat. The process is documented. And I’m already thinking about how to package this as a repeatable playbook.

Thinking about how AI could improve a high-stakes process on your team? This is exactly the kind of work I do. Reach out and let’s talk.

Not subscribed yet? Join Success & Growth.. every week, practical AI thinking for post-sales and GTM teams.

Subscribe now