Hiring is still human. Running it doesn't have to be.
Hiring is one of the most important things we do as leaders — and often one of the most poorly supported by tooling. We still rely on some of the least evolved workflows in the business: calendar ping-pong, messy note-sharing, interviews scheduled in a vacuum, and feedback cycles that stretch into weeks.
AI hasn't made hiring easier. It's made inefficiencies intolerable. It's forced us to confront just how disjointed our systems were and finally build the process we always knew we needed.
The hardest part of hiring was never the interview. It was everything around it.
What AI actually changed (the system, not the bar)
We didn't lower our standards. We still test on core skills, judgment, and most importantly problem-solving. But we changed how we run the loop:
- We can run the process ourselves without full-cycle recruiters.
- Scorecards, summaries, and feedback sync automatically.
- Everyone sees the same signal at the same time.
- Handoffs are tighter; context gets preserved.
- Candidates get far more thoughtful feedback.
AI didn't decide who to hire. It just made sure everyone was deciding based on the same information.
We've found that this alone improves decision quality dramatically. Debriefs are clearer; disagreements are more productive; and we catch red flags faster.
The best feedback happens when the interviewer is actually present. AI made that presence easier.
Treating hiring like software development
This might be the biggest unlock. We started treating the interview process the way we treat our engineering work:
- Interview questions live in GitHub. They have owners; they get updated; they go through review.
- Scorecards are versioned. We align them with what we're actually testing for.
- Interview plans are synced with our ATS. Everyone knows the goal of their round.
- Feedback is structured and additive. No more free-text blobs that say "seems smart."
We also started looking at metrics the same way we evaluate engineering teams: speed, throughput, quality, and consistency. Funnel analysis helped us identify where candidates were dropping. We tracked time to schedule, time to respond, and time to make a decision.
Once we started thinking of interviews like code, everything got easier to reason about.
Designing for the AI-native candidate
We didn't want to create an arms race against AI. We wanted to see how people use it. That meant:
- Candidates can use Claude, GPT, or any tool during live coding.
- We care less about what they remember, more about how they reason.
- We probe for how they evaluate, adapt, and verify AI-generated output.
- Most importantly — we look for whether they have a plan for AI: when to use it, when not, and why.
One of the best signals is watching what a candidate does when AI gets something wrong. Do they spot the bug? Do they prompt again? Do they panic? That tells us more than any LeetCode problem ever could.
We still care about problem-solving. AI just changed how that gets expressed. The test isn't "can you solve this alone" — it's "can you direct the system, and step in when needed."
Our actual process, step by step
Here's what the interview flow looks like today:
- Phone screen with the hiring manager. Real conversation, not checklist.
- Live technical interview. AI allowed; collaborative screen share; watch how they explore, debug, and adapt.
- Virtual coffee chat. Totally candidate-led — borrowed this from Venmo 10 years ago. This is the candidate's chance to turn the tables on us.
- Executive conversations. High-signal behavioral loops, usually with product or engineering leadership.
- References. We do them ourselves, always.
This hasn't changed much. What's changed is how cleanly we run it. Fewer gaps; less drift; faster alignment.
Candidate experience is another unlock
The biggest improvement isn't internal. It's external. Candidates get faster scheduling. They aren't asked the same question three times. They move through the loop with context and transparency.
Candidates have told us this is the most well-run process they've been through. That's not because it's fancy. It's because it's thoughtful. Clear goals; clear structure; clear feedback.
The best hiring signal isn't how candidates perform; it's how they feel when the process ends.
When you're small, every candidate touchpoint is a brand moment. This is where they decide if they trust your culture. A tight process signals seriousness and respect.
Why this works without a recruiting team
We've hired multiple people this year — without a traditional recruiting function. That's not a knock on recruiting. It's a sign of what modern tools unlock.
- AI handles logistics: scheduling, reminders, coordination.
- Interview intelligence platforms surface summaries and align feedback.
- Shared visibility means less backchanneling and fewer syncs.
What it really means is we've shifted recruiting from a high-latency function into a shared, real-time system. Hiring managers are closer to the signal. And the process doesn't stall just because one person's out for a day.
What we've learned about the people we hire
Better systems change outcomes. We've seen it firsthand.
- We spot underrated candidates earlier — people who might've been filtered out in a résumé screen.
- We make faster decisions because we're aligned earlier.
- We've reduced the number of interviews per hire without lowering the bar.
And maybe most importantly: the process has surfaced signal that goes beyond credentials. We see how people think; how they navigate ambiguity; how they use the tools at their disposal.
Closing reflection
Just like software development, hiring gets better when you tighten feedback loops, reduce coordination drag, reuse what works, and make intent explicit. That's what AI enabled us to do.
It didn't change who we hire. It changed how confidently — and respectfully — we make that decision.
If you're hiring — don't just adopt AI. Adopt the system mindset AI makes possible. Run hiring like you run product. Keep your bar high. Make the loop tight.
That's what great teams feel like on the inside — and outside.
We're building a team that works this way. If this resonates, see our open roles.
Continue the conversation on LinkedIn.
This piece is also on LinkedIn — join the discussion, or follow Matthew for the next write-up.
View the LinkedIn post