Sterling Smith

Why I'm Building RightMatch

Thu, 23 Jul 2026 00:00:00 GMT

A head of talent acquisition told me something that stuck with me. She'd post a job on a Friday. By Monday morning, she'd have hundreds of applicants waiting in her inbox. She had time to look at maybe 30 or 40 of them. The rest, likely including her best candidate, never got a real look.

I heard some version of that story over and over in 2023, and it's the reason RightMatch exists.

At the time I was a managing director at a venture fund, and before that I'd been an early employee at a startup that exited. I'd seen hiring from both sides, as the candidate and as the person building teams. So when I started digging into this, I wasn't looking for a theory, I was looking for the real bottleneck.

I talked to TA leaders and hiring managers for months. The story was always the same. Too many applicants, not enough time, and a growing sense that the best people were getting lost in the pile. Around the same time, something else was happening. Candidates started using AI to polish their resumes, so much so that by the time they showed up for a phone screen, they barely resembled the person on the page. Resumes were becoming noise, not signal.

So the first thing we built was a way to score resumes against the job description. It helped, but it didn't solve the real problem, because a well-tailored resume still isn't a real signal of how someone thinks or works. The TA leaders told us the only real fix was talking to people, but they could only interview 25 candidates a week, out of hundreds who applied.

That's the gap RightMatch was built to close. Instead of a human having to choose who gets a conversation, every candidate gets one, on day one, through an AI interview that actually evaluates how they think and what they can do. Nobody falls through the cracks anymore because nobody has to wait their turn.

That's the belief underneath everything we build at RightMatch. Hiring shouldn't be a numbers game where the best person loses because a recruiter ran out of hours in the day. It should be a fair shot for everyone who applies, and a clear signal for everyone doing the hiring.

We're still early. But I think we're standing at a real inflection point. The same forces that are making it harder to sort through applicants, more competition, more noise, more AI on both sides of the table, are also the reason a solution like this can finally exist.

My goal isn't just to fix hiring for the companies we work with today. It's to build something that changes how work gets found for a hundred million people over the next decade. That starts with one candidate, one interview, one chance that doesn't get lost in the pile.

The 25-a-week ceiling

Thu, 23 Jul 2026 00:00:00 GMT

A recruiter can interview about 25 candidates a week. That number came up over and over when I spent months talking to talent acquisition leaders. Different companies, different industries, same ceiling. Twenty-five real conversations is roughly what one human can do while also doing the rest of the job.

Now look at the other side of the funnel. A job posting goes live on Friday. By Monday there are hundreds of applicants. The recruiter has time to actually read maybe 30 or 40 of them. So the funnel looks like this: 400 apply, 40 get read, 6 get interviewed, 1 gets hired.

That means 394 people were never really considered. Not rejected. Not evaluated and found wanting. Just never seen.

Every hiring tool of the last twenty years is a coping mechanism for this ceiling. Keyword filters, knockout questions, resume scoring. They all exist to shrink hundreds down to the 25 a human can talk to. And they all share the same flaw: they select on the resume, not the person.

That flaw used to be tolerable. It isn't anymore, because candidates figured out the game. They use AI to tailor every resume to every posting, and they're right to. But it means the document your filter reads is now a mirror of your job description, not a picture of the applicant. I've watched TA leaders describe phone screens where the person on the call barely resembled the person on the page. The resume stopped being signal years ago. Most funnels just haven't admitted it.

Here's the thing: everyone in hiring optimizes around interview capacity as if it were a law of physics. It was, when interviews required human hours. It isn't now. An AI interviewer doesn't have 25 slots a week. It can give every one of the 400 applicants a real conversation on day one and surface the people who can actually do the work.

When interviewing stops being scarce, the funnel inverts. You don't sort resumes to decide who deserves a conversation. You have the conversations, then sort on what people actually said and did. Talk first. Sort after.

The recruiter's job gets better, not smaller. Twenty-five conversations a week is a brutal way to spend judgment on people a filter guessed at. Spending it on the twenty-five best performers out of 400 is what the job was supposed to be.

And the 394 finally get what they thought they were applying for: a shot.

This is the bet we're making at RightMatch. The ceiling wasn't human judgment. It was human hours. Those are different things, and we just stopped confusing them.

Why I shoot film while building AI

Thu, 23 Jul 2026 00:00:00 GMT

My day job is building AI that interviews hundreds of people in the time a recruiter used to screen one. Speed is the product. Then on weekend mornings I take a Leica M6 out into the street and shoot a machine that gives me nothing back for ten days. That's how long it takes the lab to turn a roll around. I drop it off and walk out with no idea whether a single frame worked.

I get asked about this contradiction enough that I've had to actually think about it. Here's where I landed.

A roll of Portra 400 gives you 36 frames. Every frame costs real money before you've seen it. So you slow down. You look at the light longer. You wait for the moment instead of spraying at it. You skip the shot that isn't there. Half the craft of shooting film is deciding not to press the button.

I own a digital Leica too, and I shoot it more days than not. Digital removed the constraint, and mostly that was a gift. But something got lost when every photo became free. When you can take a thousand pictures, you stop deciding which one matters. The choosing moves from before the shutter to after it, and those are not the same skill. One is attention. The other is triage.

There's no undo, either. This July 4th I pointed the M6 at fireworks. Long exposure, manual everything, no screen to check. I found out at the lab whether I got anything. You commit, and you live with the gap between the decision and the result. Most of what I do as a founder works the same way. You make the call now and the scans come back in a quarter.

I think about this in the context of what I build. AI is the biggest abundance machine ever made. Everything it touches becomes cheap and instant: words, images, code, answers. That's mostly good, the way digital photography was mostly good. But abundance is only useful if somewhere in the system a human is still paying attention, still deciding what matters, still willing to wait for the thing that's actually right.

So the two aren't a contradiction. They're the same belief pointed in opposite directions. At work, I use AI to remove the wrong scarcity: the hours in a recruiter's week that decided which applicants got seen. On a Saturday morning with the M6, film restores the right scarcity: the attention that decides what's worth a frame.

AI removes the wrong scarcity. Film restores the right one. I'd like to keep both.