Superposition Launches AI Recruiter for startup founders

We've been refining our AI recruiter with a select group of early stage founders and are excited to share what we've been working on.

I was a tech recruiter for 7 years, and have fond memories of helping Seed stage startups scale (especially placing engineer #1 at Brex). I learned to code thinking it would make me a more effective recruiter and give me greater empathy with the engineers I spoke with all day. Only the second of these proved to be true.

"Recruitment is broken" is a statement so obvious as to be trite. You've probably seen it on a dozen startup decks this morning. But it takes hands-on experience working in the recruitment industry to understand WHY it is broken.

It boils down to two perverse yet powerful incentives: commission and embarrassment.

Recruitment agencies aren't incentivised to find the perfect match for each customer. They are incentivized to find generically good candidates who are a reasonable match for all of their customers, then spray that candidate to every single role. I only got paid when one of my candidates was hired, and the constraint of time meant that I was better off finding one candidate who was a 60% match for 5 of my customers, rather than a 90% match for just one. Because the effort of getting from a 60% match ("he works at Google and went to an Ivy league") vs a 90% match (I've trawled through her github, read her Medium posts and looked up what funding stage each of her previous roles were at) was a diminishing return. Time is money when most of your work is pro bono, and so you maximize your chance of making bank by generating as many "generically good" candidates and spreading your bets by sending them to multiple roles. I made my living recruiting engineers using functional programming in San Francisco for Seed stage startups. And I was really good at it. Yet I was still subject to the same perverse incentives outlined above, and so rather than find a true Haskell evangelist for just one customer I was better off finding someone using enterprise Scala who might fancy Haskell, Elixr or OCaml and send them off to every role I was working on

And the issue of "generically good" is compounded by a fear of embarrassment that not just agencies but internal talent teams fall prey to. With the exception of a few weirdos who spend their days trawling through linkedin and their nights ploughing through SICP, recruiters aren't technical. They are ill-suited to screening technical candidates, yet we've given them the keys anyway. Their fear manifests in a job to be done of "send a candidate whose resume won't embarrass me" rather than swinging for the fences to find the best possible candidate. "No-one ever got fired for choosing AWS" is the adage in the world of buying software, and "no recruiter ever got fired for only sending resumes from AWS" is its equivalent in talent-land. And this problem is especially keenly felt when recruiting engineers, where ability to build is non-correlated with enthusiasm for perfunctory questions in a screening call. The recruiter screening call is a blunt instrument that cuts off both ends of the distribution. Cracked systems programmers who don't make eye contact are filtered out along with the scam engineers who are reading their answers off another screen. And these high upside candidates never make it past the gatekeeper to the hiring managers, who remain unaware of these false negatives.

This legacy service industry does not produce the maximally well matched candidates, so you might imagine they are differentiating based on their service levels. Perhaps this behemoth of an industry is like travel agents: I know I'm not getting access to the full market of available hotels but I'm willing to trade that off for a convenient and delightful service.

The numbers tell another story:

The US spends $204B each year on recruitment agencies, but their average NPS is worse than Workday's. Name 3 large law firms. Now do consulting companies. Then accountants. How about recruiters? I bet you struggled with that last one. There is no "McKinsey" of recruitment. There are large recruitment companies, but none of them are renowned for delivering a great service.

The high quality recruitment is done by unevenly distributed individual recruiters in niche areas. If you were hiring someone using Haskell in SF in 2018 I was your guy to call. But my knowledge was not extensible, so as soon as my customer wanted to hire someone who used Ruby, or a devops engineer or a marketer I was suddenly much less useful.

Our Vision: Revolutionizing Recruitment Through AI

This is where the vision Li and I share becomes critical. Li, a founding engineer at Vitable Health (YC S20), has firsthand experience scaling a startup from pre-seed to Series A. Like me, he's convinced that AI is poised to revolutionize vertical integration and replace repetitive intellectual jobs - with recruiting being prime for disruption. "I've always believed that repetitive intellectual jobs will eventually be automated," Li explains. "When I first heard about the AI agent approach to recruitment, I knew this was the right direction. The technology can finally address the inefficiencies I saw as an engineering leader trying to build high-performing teams." So together, we're building a multimodal AI agent that can replicate what I did well as a recruiter: chatting with companies to design their ideal hire. And a reasoning agent that can consider all the nuance of who is a good fit on paper to source candidates from across the internet at scale

So we're building a multimodal AI agent that can replicate what I did well: chatting with companies to design their ideal hire. And a reasoning agent that can consider all the nuance of who is a good fit on paper to source candidates from across the internet at scale.

Superposition starts by listening. You talk about who'd thrive at your company—not just the skills, but the vibe, the personality, and the core values. Superposition fetches examples, and we calibrate together. Then Superposition gets to work, using AI to find real matches, handle outreach, and schedule interviews right into your calendar.

Our pilot customers talk enthusiastically about the time to interviews and the quality of the matches. Superposition writes hyper personalized emails to candidates with an interested response rate 3x higher than a recruiters baseline, this gives our customers access to the top tier of talent who usually deletes generic recruiter spam.

Finding the Perfect Match: A Founder Story

I was the first customer of Superposition, and used it to make the most critical hiring match any founder has to: finding my cofounder. I needed to find a cracked engineer with exceptional design taste who shares my belief that building software is a team sport (everyone talks to customers, everyone writes code).

I used Superposition to write personalized outreach across LinkedIn, CoffeeSpace and YC's CoFounder matching. It was here that I met Li, who stood out immediately. Not only was he a founding engineer at a YC startup who had been through the journey from pre-seed to Series A, but his ability to ship incredibly fast with design sensibility was exactly what Superposition needed.

"What initially drew me to Superposition was the alignment in our vision about AI's revolutionary potential," Li shares. "But what convinced me to join as a co-founder was seeing how the product already worked - it wasn't theoretical. The technical challenge of building an AI agent that could truly understand the nuances of both companies and candidates was exactly the kind of problem I wanted to solve."

Li's technical contributions have been transformative. His experience building and scaling products has allowed us to move with extraordinary speed, turning our vision of an AI recruiter into reality. The technical architecture Li designed ensures our AI can process the complexity of human traits and job requirements while delivering matches that feel uniquely tailored to each customer.

Dogfooding Superposition allowed me to find a cofounder who was hyper matched to what Superposition needed to get to the next level

The Business and Future

We're charging our customers like a recruiter would, with a lump sum once they make a hire. There's a lot of talk about the move to outcome based pricing for AI products but the market is already used to paying recruiters based on outcomes. And the efficiencies of doing this with software not only produces a higher quality match, but provides it at fraction of the cost of a traditional recruiter

Right now we are hyper focussed on helping seed stage founders recruit founding engineers, and will grow with our customers to handle their crucial early sales, chief of staff roles and beyond.

"As a technical co-founder who's been through the startup journey," Li adds, "I know firsthand how critical those early hires are. Every wrong hire at the beginning can set you back months. What excites me most is building a product that gives founders access to talent they'd never find through traditional channels - the kind of engineers who, like me, prefer building to polishing their LinkedIn profiles."

Most founders have reluctantly accepted recruitment as an expensive, mediocre gamble. It doesn't have to be that way. We built Superposition to raise the ceiling: to deliver the kind of deeply matched, passionate candidates you'd hire yourself if you had endless hours to search.