An AI Now Charges The Recruiter's Exact Fee. What Defends The Placement When The Work Is Automated

A startup called Dex recently raised funding to build what it calls an "AI talent agent." It talks to candidates by voice or text, benchmarks their pay, prepares them for interviews, and matches them to employers through its own engine. Then it charges the employer 20 to 30% of the hired candidate's first-year salary - the identical commission model used by traditional recruitment and search agencies. Its founder, Paddy Lambros, sums up the pitch: "We only earn money if we do a good job."

This is a meaningful first. Most AI recruiting products are sold to agencies as tools. Dex is not a tool for the agency. It is priced as a replacement for one, pointed directly at the employer, charging exactly what the agency charges.

Why the matching price tag is the real story

The recruitment fee has always been defended by one implicit argument: the work is too human to automate. Sourcing the right people, screening them properly, holding the nuanced back-and-forth with candidates - these required a person, and the fee paid for that person's effort and skill. As long as the work could not be done by software, the percentage was safe.

Conversational AI changed the premise. The back-and-forth that used to require a human can now be held by a machine, at scale, around the clock. And the moment that became true, someone priced it at the agency's own rate and offered it straight to the client. The number on the invoice is the same. The only question left for the client is which version of that invoice they would rather pay.

A note on honesty

It is worth being accurate about what Dex does and does not represent. Tools like this currently concentrate in specific segments - much of the early traction is in technical and startup hiring - and it would be an overstatement to say they have taken agencies' clients at scale. The point is not that the agency is finished. The point is that the recruitment fee now has a credible, software-priced substitute for the first time, and that changes the conversation permanently. The defensible answer to "why pay you" can no longer be "because this work is too human to automate." That answer just expired.

What actually defends the fee now

If the work is no longer the justification, the justification has to be the judgement around the work - the things a matching engine, however good, cannot price.

The judgement to challenge the brief

An AI takes the role as written and finds the best match for it. A good recruiter knows when the role itself is wrong - the salary is off the market, the seniority does not fit the budget, the must-have requirement is the reason nobody good is applying. Fixing the brief is worth more than filling it, and it is the opposite of what an automated agent does.

The relationship that makes you the first call

A platform processes a transaction. It does not build the trust that makes a hiring manager phone you before they have even written the requisition, because they want your read on the market first. That relationship is not in the AI's training data and cannot be priced by it.

The read on a person no benchmark captures

Whether someone will actually thrive in a specific team, under a specific manager, in a specific culture, is a judgement built from hundreds of placements that worked and failed. It does not transfer into a model that has never had to live with a bad hire.

How to reposition the desk

• Stop selling the work. Any pitch built on "we do the sourcing and screening so you do not have to" now describes a piece of software with the same price tag. Retire it.
• Sell the judgement. Make the advisory layer - challenging the brief, advising on the hire, reading the fit - the explicit, central thing the client is paying for.
• Use the tools to free up time for it. Let AI do the sourcing grind, and reinvest the hours into the relationship and judgement that justify the fee.
• Concentrate on the roles where judgement dominates. The further a hire is from a cleanly specified, easily matched role, the weaker the AI's advantage and the stronger yours.

The bottom line

For twenty years the recruitment fee was defended by the difficulty of the work. Dex, and the wave it represents, ends that defence by automating the work and charging the same rate for it. The fee does not disappear - but it now has to be defended by the judgement around the work rather than the work itself. The recruiters who make that shift, and make it before the substitute reaches their market, will keep a fee worth charging. The ones who keep selling the work will find themselves competing on price with software that never sleeps.