The best AI recruitment agencies, compared

How the four were judged

Every agency claims a network and a process. Five things separate them in practice, and each row of the table above maps to one.

• Specialism, in placements rather than in slogans. An agency that placed forty software engineers and two AI researchers last year is a software agency. Ask for the split.
• Where the candidates come from. A job board and a CV database see the people who are looking. The published literature, conference programmes and public code see everyone who does the work, including the majority who are not looking.
• How the fee is structured. A retainer pays for activity. A success fee pays for a hire. The structure tells you who carries the risk of a search that fails.
• Time to a first shortlist. A specialist with a database or an index should show names within two weeks. Longer means the search starts from zero.
• Where they can hire. AI research is concentrated: the top two countries hold 54% of the researchers with a known country in the index. An agency that hires in one country works from a fraction of the supply.

1. Founding Hires

Founding Hires is a recruitment agency for AI researchers, applied AI engineers and AI engineers. It is the firm behind this page, so read the next paragraphs as a claim to be checked, and check it against the method page and the atlas, which publish the numbers the claim rests on.

Three things set the firm apart. The first is sector expertise: the agency only recruits for AI roles, and every search starts by reading what the candidate has published or shipped rather than a CV. The second is the index. Founding Hires maintains its own index of the AI research literature, 614 881 researchers on 291 498 papers, the largest index of AI researchers we know of, and for 47 289 of them a public source dates the end of the doctorate. That date is what lets the agency approach a researcher six to nine months before they come on the market, when a first message still gets read. The third is the founder, who has worked alongside some of the best-known AI companies in the world and runs each search personally.

The fee is a success fee only. There is no retainer, and nothing is owed until someone signs. A first shortlist arrives within seven days of the intake call, with the reasoning attached: what each person has built, why they fit and where the risk is. Clients who would rather run the search themselves can access the index directly and set alerts on the subjects they hire on.

Where it fits less well: a company that needs twenty engineers by the end of the quarter, or contractors on day rates. The firm runs a few searches at a time and says no to the rest.

2. Flaire

Flaire is a small, highly specialised AI recruitment firm whose network is built around Europe's best-known AI companies. Where a large group sells reach, Flaire sells proximity: the recruiters know the teams they place into, and the introductions carry that context.

It fits a company that wants a boutique, values a warm introduction over a long list, and hires in Europe. It fits less well when the role needs a candidate who has never touched the European AI scene, or when the search has to run across several countries at once. Fees are not published; ask for the model and the rebate period before the intake call.

3. Tenth Revolution Group

Tenth Revolution Group is a large technology recruitment group operating through several brands, with the recruiter headcount and the candidate database that come with that size. In AI, its strongest suit is the forward deployed engineer: the person a vendor places inside a customer to make a model work on that customer's data, a profile that mixes engineering, delivery and a tolerance for other people's infrastructure.

It fits a company that needs many hires, needs contractors, or needs a role filled in a country where it has no network of its own. It fits less well for a single research hire where the candidate is not on any database and the search has to start from a paper. Fees are not published; contract placements carry a margin on the day rate, and permanent placements a share of first-year salary.

4. Harnham

Harnham is a specialist in data and analytics recruitment, covering data science, machine learning engineering and analytics roles, with offices in the United Kingdom, the United States and Europe. It works from a large database in its field and places both contract and permanent staff.

It fits a company building or scaling a data team, where the roles are defined and the candidates describe themselves in the same words the job does. It fits less well for a frontier research role, where the strongest candidates are doctoral students who do not yet call themselves anything. Fees are not published; ask.

How to choose an AI recruitment agency

Start from the role, not from the agency. The four firms above serve different problems, and the wrong one will work hard and deliver the wrong people.

• A researcher, or an engineer who was one. The candidate is inside a university or has just left one, publishes, and is not on the market. You need an agency that reads the literature and knows when doctorates end. Founding Hires is built for this search.
• An applied AI engineer with a defined stack. The candidate moved from research into industry in the last two years and is reachable for a window of twelve to thirty months. Both a literature index and a strong database find these people; ask each agency how it dates the move.
• Several engineers, or contractors. Volume and speed favour a large group with recruiter capacity. Ask how many searches each recruiter runs at once.
• A data team. The roles are standard and the candidates are numerous. A data and analytics specialist such as Harnham has the database.
• A European boutique search. One senior hire, introductions that carry weight, and a preference for a small firm. Flaire.

What an AI recruiting agency costs

Four fee models are in use, and the model matters more than the percentage because it decides who carries the risk of a search that fails.

• Retained search. A fee paid in instalments, typically a third at the start, a third at shortlist and a third at offer. The agency is paid for the work whether or not anyone signs. Common for executive roles.
• Contingency. A share of first-year salary, owed when a candidate the agency introduced signs. Nothing is owed otherwise, so agencies on contingency run many searches at once and give each one the time it earns.
• Success only, with an exclusive search. The fee is owed on a signed offer, as with contingency, but the agency runs the search exclusively and to completion. Founding Hires works this way: no retainer, and the risk of the search sits with the agency until someone joins.
• Contract staffing. A margin on the contractor's day rate for as long as the contract runs. Cheap to start and expensive to keep.

Whatever the model, three terms belong in writing before the search starts: the share or the fee, the rebate if the hire leaves within the first months, and whether the search is exclusive.

Questions to ask before you sign

• Where will the candidates come from, and can you show me? An agency that works from a source can show the source. One that says "our network" cannot.
• Which of your placements in the last year were AI researchers or AI engineers? Names are confidential; the split is not.
• Who runs the search? The person on the sales call and the person who will read the papers are often different people.
• How many searches does that person run at once? Above five or six, yours gets the time that is left.
• What happens when the role as written cannot be filled? The right answer is that the agency says so early and tells you why. The wrong answer is that it keeps sending people.

When you do not need an agency

An agency earns its fee where the supply is hard to see. When it is not, run the search yourself. Senior engineers with a public GitHub record can be found and approached directly, and how to find and hire AI engineers sets out the method. A single research hire where you already know the three labs that matter is a week of reading, and how to find and hire AI researchers covers the reading. If the open question is which profile the problem calls for, start with AI researcher, applied AI engineer or AI product engineer. The index that the agency searches is also open to clients directly, with alerts on the subjects you hire on.

Common questions

What is the best recruitment agency for AI researchers?

The one that finds researchers where they are, in the published literature, rather than on job boards. Most AI researchers are inside universities and labs and never post a CV. Founding Hires searches its own index of the AI research literature and is paid on success. A boutique such as Flaire fits a small search built on a European network. A large group such as Tenth Revolution Group fits volume hiring and contract staff.

How much does an AI recruiting agency charge?

Three models exist. Retained search takes a fee in instalments from the start of the search. Contingency charges a share of first-year salary when someone is placed, with nothing owed otherwise. Success-only search, which is how Founding Hires works, charges only on a signed offer and takes no retainer. Contract staffing adds a margin to a day rate instead. Ask for the model, the share and the rebate period in writing.

What is the difference between an AI recruitment agency and a generalist tech recruiter?

A generalist matches keywords on CVs. An AI recruitment agency can read a paper or a repository, tell a first author from a fourth, know which venues carry weight, and reach people who are not looking. The test is simple: ask the recruiter to explain what your last hire published and why it mattered.

Should we use an agency to hire applied AI engineers and AI engineers, or only for researchers?

An agency earns its fee where the supply is hard to see. Researchers finishing a doctorate and engineers who moved from research into industry within the last two years are both hard to see from a job board, so an agency helps for both. Senior engineers with a public GitHub record are often reachable directly, and the how-to guides on this site cover that search.

How long does an AI recruitment agency take to deliver candidates?

A specialist with a database or an index delivers a first shortlist within one to two weeks. Founding Hires works to a first shortlist in seven days. The time to a signed offer then depends on the candidate's window: a researcher finishing in September is reachable from January and decided by May, and no agency changes that calendar.

What should we ask an AI recruitment agency before signing?

Where the candidates come from, and whether the agency can show the source. How many searches it runs at once per recruiter. Which of its placements in the last year were AI researchers or AI engineers rather than general software roles. The fee model, the share, the rebate period and the exclusivity terms. And who exactly will run the search.