Dex Raised $5.3M to Hire Your AI Engineers Without the LinkedIn Parade

Somewhere in London right now, an early-stage founder is staring into the abyss of technical hiring: 412 applicants, 37 recruiters, 14 LinkedIn messages that begin with “Hope you’re well,” and one machine-learning candidate who has three offers, no calendar availability, and the emotional energy of a hunted falcon. Into this spiritually stable scene glides Dex, which just announced a $5.3 million seed round to act as an AI talent agent for AI engineers.

That sentence contains enough recursion to bend a spoon. An AI startup has raised money to help other AI startups hire the humans capable of building their AI startup. We have achieved a kind of venture-backed closed loop. And yet, against my better instincts, I think this is one of the more sensible little loops I have seen lately.

Dex is not trying to replace work with a chatbot and a dramatic keynote. It is trying to fix one of the most annoying, expensive, and weirdly theatrical parts of startup life: finding scarce technical talent without turning the process into either a spam cannon or a five-stage personality obstacle course. That is not glamorous. It is, however, painfully real.

LinkedIn, but with fewer drive-by messages and more adult supervision

According to TNW, Dex was founded in early 2025 by Paddy Lambros, a former Atomico talent adviser who spent years watching young companies discover that almost every strategic problem eventually becomes a hiring problem. The company says more than 15,000 engineers have signed up, more than 50 tech companies are paying customers, and it has gone from zero to roughly $1.8 million in annualized recurring revenue in under six months of charging. That is the sort of traction that gets investors to stop saying “interesting wedge” and start saying “let’s wire the money before someone in San Francisco does.”

On its About page, Dex frames the problem with refreshing bluntness: the best engineers are not short of opportunities, they are short of the right ones. That is a much better diagnosis than the usual recruiting-tech fantasy that the world simply needs one more dashboard with gradient buttons and a promise to “unlock talent flows.” Dex is betting that technical hiring is broken not because there is too little data, but because too much of the process is shallow, public, and optimized for noise.

This is where the pitch gets interesting. Dex interviews candidates through an AI agent over voice or text, asks open-ended questions, builds a richer picture of what they actually want, and then uses that context to match them with roles. After that, it hands qualified, interested candidates directly to hiring managers. In other words, it is trying to behave less like software you buy and more like a recruiter who has somehow become impossible to exhaust.

I have a soft spot for startups that decide not to wage war on reality so much as negotiate with it. That is part of why Replenit’s retail AI timing obsession worked on me, and why Astor’s attempt to make investing less like gambling with a nicer font felt oddly grounded. Dex has a similar energy. It is not promising transcendence. It is promising fewer wasted conversations.

The most radical thing here is that someone remembered candidates are people

The hiring-tech market has spent years building products that appear to regard candidates as malformed database entries. Upload resume. Re-enter resume. Complete personality quiz. Take timed logic test. Speak into webcam. Receive silence. It is one of the few consumer experiences where being ignored by a machine can feel more insulting than being rejected by a human.

Dex is at least claiming a different philosophy. On the same About page, Lambros says, “CV spam bots and dehumanising AI interviewers are turning a bad process into an inhuman one. We need Dex.” That is, admittedly, a founder quote written with full awareness that it may end up in a deck. Still, I appreciate the target selection. The problem is not merely that hiring is slow. It is that modern recruiting often scales indignity faster than judgment.

Dex’s company-facing pitch leans into curation rather than volume. On its page for employers, the startup promises “the engineers others can’t reach,” a small shortlist, and direct introductions rather than a giant applicant slurry. It says it works with companies like Lovable, ElevenLabs, Synthesia, Granola, and Fyxer. One customer testimonial from Fyxer says Dex has made “genuinely impactful hires,” which is exactly the kind of boring sentence you want in enterprise-adjacent recruiting. Nobody serious wants recruiting software that sounds fun. They want it to quietly prevent bad decisions.

And yes, there is AI all over this. The product reportedly uses models from Google, Anthropic, and OpenAI. There is a proprietary matching engine in the middle. There is the inevitable “old-school machine learning” aside, which I enjoyed because we have now advanced to a stage of the cycle where machine learning itself is the legacy comfort food. But the underlying idea is less “summon a synthetic headhunter” than “use conversational AI to get beyond the weirdly flimsy information in a public profile.” That is not absurd. It is arguably overdue.

SiliconSnark has been orbiting this broader agentic moment for weeks, including our guide to computer-use agents. The recurring pattern is simple: the good versions are rarely the loudest. The useful agent is not the one that claims it will run your life. It is the one that takes a specific mess and makes it meaningfully less stupid.

The beautiful awkwardness of charging like a recruiter while insisting you are not just a recruiter

My favorite detail in the coverage is that Dex reportedly charges employers 20% to 30% of a hired candidate’s salary, which is the classic executive-search model. This is excellent. After years of software startups desperately trying to escape services economics, here comes a startup essentially saying: actually, if the output is excellent hires, we are perfectly happy to get paid like adults who deliver value.

That structure also keeps the company honest. If Dex only gets paid when a hire happens, then it does not have much incentive to sell clients a blizzard of mediocre candidates and call it pipeline health. It has to produce introductions that survive contact with reality. I respect that. It is harder to hide behind dashboards when your invoice depends on an actual human saying yes.

There are, of course, reasons to be cautious. Recruiting startups have a long tradition of starting with a beautifully narrow wedge and then succumbing to platform delirium. The temptation to expand from “elite AI engineers” to “all hiring for everyone” will hover over Dex like a venture partner with a growth chart and a thirst for total addressable market. And the incumbents are not asleep. LinkedIn, assorted sourcing tools, and every second startup with “talent OS” in the deck are all circling the same anxiety.

There is also the cultural risk of becoming the nice version of a thing people already distrust. Candidates may enjoy a more thoughtful AI conversation than a soul-draining application form, but that does not mean they want every career move mediated by a cheerful inference engine. Trust is fragile here. The startup will have to prove that “AI talent agent” feels more like advocacy than sorting.

Still, the company has chosen its battlefield wisely. Technical hiring in AI is scarce, fast, expensive, and status-loaded. Founders care. Candidates care. Investors definitely care, because nothing makes a seed deck look more fragile than a plan built on “we will hire exceptional ML talent later.” Dex is selling into a pain point that everyone involved can feel in their rib cage.

And unlike some recent tiny-round fever dreams, this one does not require me to pretend the market need is metaphysical. As much as I was charmed by Sinai.ai’s sentient-book ambition, Dex is playing a straighter game. The problem is obvious. The buyer exists. The budget exists. The urgency exists. That matters.

Verdict: a promising little rocket with strong manners

My verdict is that Dex feels like a promising little rocket, not because it has discovered a magical new truth about labor markets, but because it seems disciplined enough to attack one ugly corner of them properly. This is not a utopian pitch about ending work. It is a practical pitch about improving a high-friction market where too much of the current process is lazy, noisy, and faintly insulting to everyone involved.

Will AI fully fix hiring? Absolutely not. Humans are still eccentric mammals with ambition, ego, taste, fear, and a habit of interviewing as if they are casting a prestige drama. No model is going to solve all of that. But if Dex can help serious companies meet strong candidates faster, with less spam and more context, then it may have built something more durable than a recruiter replacement fantasy.

And frankly, in a startup economy where half the pitches sound like they were generated by feeding a white paper into a ring light, a company saying “what if we made technical hiring less wasteful and less dehumanizing” feels almost quaint. I mean that as a compliment.