The best AI recruiting tools, and what the law still lets you do with them

The best recruiting tool with artificial intelligence depends first on the size of your team. For a French small company that wants an applicant tracking system and hosting in France, Taleez is the most direct choice. For a fast-growing company that wants a price known upfront, Ashby is the only one of the eight that publishes one. For pure sourcing, LinkedIn's agent sits where the candidates are. And if what you need is an AI brick to plug under an existing tool, HrFlow.ai does that job out of France.

Here are the eight AI recruiting tools, their prices as taken from vendor pages in September 2026 where they publish any, and above all the regulatory question almost every comparison in this category gets wrong.

How this comparison was built

Five criteria, stated before the first name. They are few because a fifteen-column comparison is unreadable and decides nothing.

• The real price, taken from the vendor's page and dated to the month. Where there is none, it is stated, and that is information in itself.
• What the AI actually does, and at which stage. Screening CVs, finding profiles and writing up an interview are three different jobs.
• The underlying model, where the vendor publishes it. In this category almost none do, and that has to be said.
• Hosting and vendor identity, checked on the pricing or trust page, never on a flag pasted onto a homepage.
• Where human decision sits, because recruiting is the one area on this blog where European regulation already forces an answer.

On transparency, here is the exact position, and it is uncomfortable. Tandem has not deployed any of these eight tools at a client. Our first-hand material on the subject comes from elsewhere: our own AI Builder hiring, the sourcing I do on LinkedIn for those roles, our daily use of AI note-takers, and the voice pre-qualification agents Tandem deploys on engagements. Every card says so. For method and use cases on the HR side, that work lives in our article on AI agents for HR and recruitment, and this comparison does not rewrite it.

The eight AI recruiting tools in one table

1. Taleez, the French tool for smaller companies

Taleez is an applicant tracking system built in France and unapologetic about it, with an explicit focus on small and mid-sized companies. The product covers job distribution, application management, evaluation and reporting. Its AI layer stays modest: CV matching against open roles, and automatic application analysis that surfaces the relevant profiles.

Key features

• Job distribution is built in, which avoids yet another tool just to publish on job boards.
• CV matching automatically identifies the applications that fit an open role.
• Automatic application analysis ranks the files to handle without the recruiter opening every attachment.
• Recruiting analytics cover time to hire, sources and stage-by-stage conversion.
• Support is in French, which weighs more on HR tool adoption than people expect.

Underlying model and limits

Taleez does not publish the model behind its AI layer, and that is true of seven vendors out of eight in this comparison. The vendor does state that data is hosted in France, which makes it the only one here on that point. The limit is functional ambition. Taleez screens and ranks; it does not source candidates for you and does not automate interviews.

Pricing

No price is published. The pricing page explains that the quote is built on organisation size, hiring volumes, specific needs and company type, with a promise of no hidden fees mid-contract. As of September 2026, you have to go through a sales conversation to learn the amount.

What users say

Taleez displays a Trustpilot score of 4.8 out of 5 on its own homepage, as of September 2026. On Capterra it scores 5 out of 5, but on five reviews only, which makes it unreadable. That is the classic situation of French vendors in this category: real local reputation, and an international review base too small to interpret.

My take on Taleez

I have not deployed it on an engagement, so I stick to what is verifiable. What I do observe in audits is that the hosting-country question comes up every single time a legal department joins the project, and that very few tools in this category can answer it in one sentence. Taleez can. In exchange, do not expect Ashby's analytical depth or LinkedIn's sourcing power. It is a good foundation, not a platform.

2. Welcome to the Jungle, employer brand first

Welcome to the Jungle started as a jobs media outlet that became a software vendor, and that lineage explains its whole positioning. Where others start from the internal process, the Welcome Hiring Suite starts from attractiveness, with editorialised company pages, studio-produced content, then an applicant tracking system and a matching layer. The vendor claims more than 4,500 corporate customers and over 500,000 registered talents.

Key features

• The editorialised company page remains the entry product, and it lives on a media outlet candidates already read.
• Welcome Job Matching connects open roles to the talent base registered on the platform.
• The applicant tracking system is presented as designed around AI rather than retrofitted.
• Sourcing draws on a candidate base already qualified by the act of registering with the media outlet.
• An in-house studio produces employer brand photography and video, as a paid option.

Underlying model and limits

The vendor calls its applicant tracking system AI-native without detailing what that covers, and without naming a model. The limit is twofold. Pricing depends on hiring volume and stacked modules, so it is hard to anticipate. And the tool's value is tied to the media outlet's audience, which makes it very strong on tech and marketing roles in France, and much less so on industrial profiles in the regions.

Pricing

No price is published. The offer page advertises pricing adapted to your hiring volume, a bundled subscription for the suite, role packs sold as credits, and employer brand content as an extra. As of September 2026.

What users say

Welcome to the Jungle has no usable review base on software marketplaces for its vendor side, most public reviews covering the media outlet from the candidate's point of view. That is stated as it is, rather than padding with a score that would not measure the right product.

My take on Welcome to the Jungle

I have not deployed it on an engagement. I do, however, hire AI Builders at Tandem, and I see which way the applications come in. On tech profiles around Paris, the media outlet's audience makes a real difference, and no AI layer compensates for a job ad nobody reads. That is the blind spot of comparisons that only assess features. On roles the outlet covers less well, the trade-off flips.

3. Teamtailor, the polished candidate experience

Teamtailor is a Swedish vendor that built its reputation on the quality of the candidate journey, at a time when most recruiting software cared mainly about the recruiter's dashboard. The product covers career pages, applicant tracking, follow-up automation and an AI layer called AI Co-pilot. It is widely used in France among mid-sized companies.

Key features

• Career pages are built without a developer and remain the product's historic strength.
• The AI Co-pilot takes on the recruiter's repetitive tasks, from writing the ad to initial screening.
• Follow-up automation prevents applications left unanswered, the top source of candidate dissatisfaction.
• Collaborative review lets hiring managers assess candidates without prior training on the tool.
• The integration ecosystem covers the main payroll and human resources management tools.

Underlying model and limits

Teamtailor names no model and does not detail its AI layer on its pricing page, where a single label appears. The vendor is registered in Stockholm, but the page does not say where data lives, which calls for a contractual check before committing. The functional limit is analytics, shallower than Ashby's, and outbound sourcing, less equipped than LinkedIn's.

Pricing

No price is published. The pricing page simply invites you to request a quote, with no grid and no indicative tier, as of September 2026. It is the only one of the eight giving no clue at all about its pricing structure, neither per seat, nor per volume, nor by company size.

What users say

Teamtailor scores 4.6 out of 5 on around 505 G2 reviews in September 2026, the largest review base in this comparison after LinkedIn. The recurring compliment is ease of adoption and the quality of the career pages. The recurring complaint is reporting depth and cost relative to the actual scope.

My take on Teamtailor

I have not deployed it on an engagement, so I stick to the verifiable. My overall view is to look at it for what it does best, candidate experience, and not for its AI layer, which is the least documented of the eight. If your problem is a high application drop-off rate, it is a very strong candidate. If your problem is finding rare profiles, Teamtailor's AI will not solve it.

4. SmartRecruiters, the enterprise player now inside SAP

SmartRecruiters is a hiring platform for large organisations, founded by a French entrepreneur and long independent. That is no longer the case. SAP completed its acquisition on 11 September 2025, after announcing it on 1 August of the same year. The vendor continues to operate as a standalone business inside the group, with integration into the SAP SuccessFactors suite announced. Its AI layer is called Winston.

Key features

• Winston Companion absorbs repetitive work, from initial screening to interview scheduling.
• Winston Match identifies the best candidates for a role and offers suggestions unprompted.
• Winston Chat guides the candidate through the application and personalises recommended roles.
• The platform covers applicant tracking, candidate relationship management and onboarding.
• Job distribution and dynamic scheduling are included from the first tier.

Underlying model and limits

SmartRecruiters is the only one of the eight to give a technical hint, describing Winston Chat as relying on advanced large multimodal models, without naming which. The limit is total cost and horizon. A rollout needs an integrator and several months, and the SAP acquisition adds a roadmap uncertainty that has to be put on the table before signing a three-year contract.

Pricing

The pricing page shows an Essential tier starting at 14,995 dollars, around 13,800 euros, without stating the period, and sends the other three tiers to a quote. As of September 2026. It is the only public figure anywhere in the high end of this category.

What users say

SmartRecruiters scores 3.9 out of 5 on G2 in September 2026, the lowest score in this comparison, but on a review base that is very small relative to its market share, which calls for caution. The recurring compliment is the ability to absorb complex, multi-country processes. The recurring complaint is the weight of the tool and the dependence on a dedicated administrator.

My take on SmartRecruiters

I have not deployed it at a client, so I stick to what documentation and reviews make verifiable. What I would say in a framing session is that the SAP acquisition is the only piece of information that really matters here. If your company already runs on SuccessFactors, the tie-up works in your favour. If you are not in the SAP world, you are buying a product whose roadmap will be decided elsewhere.

5. Ashby, the only one that shows its price

Ashby is a hiring platform born in the high-growth technology ecosystem, and its reputation was made on analytics. Where most tools in this category produce decorative reports, Ashby lets you measure a hiring funnel the way you measure a sales funnel. The customer logos on its homepage say which segment it targets: Shopify, Notion, Linear, Ramp, Vanta and Zapier.

Key features

• Funnel analytics is the product's real strength, with reports crossing source, stage and time.
• Applicant tracking, scheduling and sourcing live in the same tool, with no add-on module.
• AI credits are explicit and counted, which makes consumption readable rather than magical.
• Ashby Analytics plugs into existing recruiting tools without replacing what is in place.
• Specific pricing is offered for emerging markets, which is rare in this category.

Underlying model and limits

Ashby names no model, but it is the only one to quantify AI usage, with 1,500 credits a month on the first tier, then 2,500 and 12,500 credits per seat per year on the ones above. The limit is segment. The first tier stops at a hundred employees, and beyond that the price returns to a quote. The tool is also very Anglo-Saxon in its conventions, which takes adaptation work on French hiring processes.

Pricing

Ashby's pricing page shows 400 dollars a month, around 370 euros, for the Foundations tier up to a hundred employees, with a 10 % discount on annual billing. The Plus and Enterprise tiers are quote-only. As of September 2026. It is the only clear monthly price among the eight tools in this comparison.

What users say

Ashby scores 4.7 out of 5 on around 130 G2 reviews in September 2026, the second-best score in this comparison. The recurring compliment is the depth of reporting and the speed of the interface. The recurring complaint is the initial learning curve and a high price for a small team.

My take on Ashby

I have not deployed it on an engagement. What I take from it is that pricing transparency is itself a selection criterion in this category, and Ashby is the only one that respects it. Counting AI credits rather than promising unlimited AI is the same instinct for seriousness. The real test before signing is to ask for the conversion: how many credits does screening a hundred applications consume. No vendor in this comparison publishes that table.

6. LinkedIn Hiring Assistant, the agent where the candidates are

Hiring Assistant is LinkedIn's first AI agent, generally available since late September 2025 and now offered in seven languages including French. It does not replace an applicant tracking system; it attacks the step before, sourcing. Its product page describes it as building a search strategy from a conversation with the recruiter, then running dozens of searches and learning from the feedback it gets.

Key features

• The agent turns a conversation with the recruiter into a search strategy, then executes it across the network.
• It learns from the recruiter's approvals and rejections, and refines subsequent searches.
• It writes and sends the outreach, with automated follow-ups.
• It draws on network signals, which no tool outside LinkedIn can reproduce.
• It works inside the existing LinkedIn Recruiter flow, with no new tool to install.

Underlying model and limits

LinkedIn does not publish the model behind the agent, which is consistent with its parent company's practice on this kind of brick. The figures shown come from the vendor: 81 % fewer profiles to review, 66 % higher InMail acceptance and an hour and a half saved per role. One named customer, Expedia Group, reports thirty days cut from time to hire. The limit is structural. The agent only sees LinkedIn, it will tell you nothing about a candidate who is not there, and it locks you into a single channel.

Pricing

No price is published. Hiring Assistant is sold as an add-on to LinkedIn Recruiter, and goes through the Recruiter Professional Services Plus offer for staffing agencies, as of September 2026. The real cost therefore depends on your existing Recruiter contract, which makes any comparison with the other tools in this list impossible without a quote.

What users say

LinkedIn Recruiter, the product that carries the agent, scores 4.5 out of 5 on around 420 G2 reviews in September 2026. The recurring compliment is the unmatched depth of the profile base. The recurring complaint is price, and the saturation of the most-contacted candidates, an effect that automating sourcing can only worsen.

My take on LinkedIn Hiring Assistant

This is the card I have the most lived experience with, because I source on LinkedIn myself for our AI Builder roles at Tandem. What I observe is that the hard part is never finding profiles, it is writing a message a good candidate answers. An agent that sends more messages faster improves one metric and degrades the other. I explained in my post on the AI Builder role taking off why those profiles have become scarce, and scarcity is not solved by volume. Worth testing on one campaign, not generalising from the start.

7. HrFlow.ai, the French brick under the tool

HrFlow.ai is a French vendor that does not sell recruiting software but the AI infrastructure that plugs underneath it. That is a category of its own in this comparison, and it is exactly what makes it interesting for a company that already has a tool and does not want to change it. The vendor claims more than a thousand customers: HR software vendors, staffing agencies and large employers.

Key features

• The parsing API turns a CV into usable structured data, whatever its original format.
• The scoring API matches a profile against a role and returns a score, with the criteria behind it.
• The tagging and embedding APIs power semantic search across an existing application base.
• The upskilling API identifies a profile's adjacent skills, which is useful for internal mobility.
• Everything is exposed as a programming interface, so it plugs into a custom workflow.

Underlying model and limits

HrFlow.ai advertises foundation models trained on recruiting data, and publishes a comparative accuracy chart on its homepage, without those measurements being verifiable by a third party. The vendor states it hosts on AWS in Ireland and Google in Belgium, so inside the European Union. The limit is obvious and owned: there is no recruiter interface. Without a technical team or an integration partner, this tool is not usable.

Pricing

Billing depends on the API. Parsing is charged per successful request, scoring on active records, meaning those neither archived nor deleted. The vendor offers a free tier of a thousand requests a month plus pay-as-you-go or annual subscription plans, but publishes no unit price. As of September 2026.

What users say

HrFlow.ai has no usable review volume on software marketplaces, which is consistent with a product sold to engineering teams rather than to recruiters. That is stated as it is, rather than presenting a score built on a handful of reviews as a signal.

My take on HrFlow.ai

I have not deployed it at a client, but it is the brick whose architecture most resembles what Tandem builds on engagements. When we automate document processing, we assemble an orchestration tool, an extraction brick and a small model, rather than one piece of software claiming to do everything. HrFlow.ai is exactly the extraction and scoring piece of that assembly, on the HR side, and the fact that it is French and hosted in Europe counts. The real question is not its price, it is whether you have someone to integrate it. If the answer is no, take a tool from the list above, and read our article on AI agents for HR instead.

8. Metaview, from note-taking to agents

Metaview is a London company founded in 2018 by Siadhal Magos and Shahriar Tajbakhsh, both formerly of Uber and Palantir. The tool made its name as an interview note-taker, and that is still what it does best. It has since widened into an agent platform covering sourcing, application review, screening, notes and reports. The vendor claims more than 2,500 customers, including Sony, Deel, ElevenLabs and Deliveroo, and raised 35 million dollars in a Series B led by GV.

Key features

• The note-taker records, transcribes and summarises an interview, and fills in the scorecard.
• The sourcing agent looks for profiles from an instruction written in plain language.
• Automatic application review ranks incoming files before the recruiter steps in.
• Write-ups flow into the applicant tracking system, through a wide integration catalogue.
• Reports aggregate what was actually said in interviews, not only what was ticked.

Underlying model and limits

Metaview describes a context-aware AI able to handle unstructured data, without naming a model or provider on its pricing page. The limit is the one shared by every note-taker, and public reviews confirm it. Transcription and scorecard accuracy remain uneven, customisation is limited, and the tool struggles with some languages and accents. For an interview held in French with a non-French-speaking candidate, that is a real issue.

Pricing

Metaview's pricing page shows, on its sourcing agent, a free tier up to a hundred profiles, a Pro tier at 100 dollars per user per month, around 92 euros, for two hundred profiles, and a Max tier at 300 dollars, around 275 euros, for unlimited. The full platform is quote-only. As of September 2026.

What users say

Metaview scores 4.8 out of 5 on around 130 G2 reviews in September 2026, the best score in this comparison. The recurring compliment is ease of use and the time saved by no longer taking notes. The recurring complaint is transcription and scorecard accuracy, customisation limits, and integration gaps.

My take on Metaview

I have not deployed it at a client, but Tandem uses AI note-takers every day, and I know exactly where the value sits. It is not in the summary, which nobody rereads. It is in the fact that the information lands in the right place with no re-entry. That is the same lesson as on sales tools. We compared the main note-takers in our roundup of the best AI meeting note-takers, and what separates them is never the quality of the text produced, it is the quality of the sync.

Which AI recruiting tool should you pick for your situation

• You hire fewer than twenty people a year, in France. Taleez, for hosting in France, French-language support and a scope that does not overwhelm you.
• Your problem is attracting, not screening. Welcome to the Jungle, whose value comes from the media outlet's audience as much as from the software.
• Your candidates drop out mid-application. Teamtailor, which built its product on exactly that problem.
• You want to measure your funnel seriously. Ashby, and its published price will save you three sales meetings.
• You are hunting rare profiles and you live on LinkedIn. Hiring Assistant, as an add-on to your existing Recruiter contract.
• You run many interviews and you lose the information. Metaview, starting with its free tier to check quality in your own language.
• You already have a tool and an engineering team. HrFlow.ai, to add parsing and scoring without changing software.
• You hire across several countries with heavy processes. SmartRecruiters, asking the SAP question before you sign.

What nobody says about AI recruiting tools

The AI Act timeline moved, and almost no comparison noticed

CV screening, candidate scoring and automated interviewing are high-risk uses under Annex III of the European regulation on artificial intelligence. On that basis, most articles published this year explain that the heavy obligations, technical documentation, incident registry, formalised human oversight, have applied since 2 August 2026. That has been wrong since the summer.

The Council of the European Union gave its final green light on 29 June 2026 to a simplification regulation that pushes those deadlines back, after a provisional agreement with Parliament on 7 May. The new dates are 2 December 2027 for stand-alone high-risk systems, which covers all eight tools in this comparison, and 2 August 2028 for those embedded in a product. Prohibited practices, including emotion recognition on a candidate, have been banned since 2 February 2025.

Telling the candidate is not optional, and it already applies

While the European deadline recedes, a French obligation applies right now. The Labour Code requires that a candidate be informed, before they are used, of the recruitment-aid methods and techniques applied to them, and requires those methods to be relevant to the purpose pursued. The French data protection authority has published a guide for recruiters recommending that this information be given in writing and in advance, individually or collectively. Silent automated screening is therefore not compliant today, regardless of the AI Act.

The practical consequence is easy to test in a demo. Ask to see where the tool shows that notice to the candidate, and who can switch it off. Of the eight tools in this comparison, none puts that answer forward on its website.

The real bottleneck is not screening, it is pre-qualification

Every vendor in this category sells application screening, because it is visible and easy to demo. That is not where the time goes. It goes into the first exchanges, those fifteen-minute calls that exist to check three facts a form failed to capture. I explained in my edition on voice agents why this is recruiting's next turn. The voice agent replaces those first qualification calls, asks everyone the same questions, structures the answers and feeds the HR tools behind it.

The counter-intuitive argument is consistency. An AI does not tire on the tenth call of the day, does not get irritated by a confused answer and keeps the same tone from the first candidate to the last. On a high-volume hiring campaign, that regularity is a fairness gain as much as a time gain, and very few articles on the subject acknowledge it.

One last point, and it puts the subject back in proportion. The cost of a hire is not limited to the recruiter's time. I wrote in my review of the past year that once you count recruiting, onboarding, management and turnover, the trade-off between hiring and automating changes shape entirely. That applies to the profiles you are looking for too. The AI Builder role I described in my edition on the return on investment of AI has become hard to hire precisely because hundreds of companies are looking for it at the same time. No tool in this list solves that.

So which AI recruiting tool should you choose in 2026?

If you were to keep three names, here are mine. Taleez for a French small company that wants a clean foundation, support in its own language and data hosted in France. Ashby for a growing company that wants to measure its funnel and know its price before the sales meeting. Metaview for a team running many interviews and losing information between the conversation and the decision, testing its free tier first on interviews in your own language.

But the tool choice will not decide your outcome. What will decide it is the quality of your job ads and how fast you answer candidates. A middling tool on a fast process beats an excellent tool on a process that leaves a good profile unanswered for three weeks. Start by measuring your time between application and first contact. That figure depends on no vendor, and it is the one losing you the best people.

Tandem supports teams through that framing, from tool selection to wiring the agents. If you want to know what can be automated at your company before buying anything, start with an AI audit of your processes. If the framing is done and you are looking for delivery, our Autopilote offer covers building and running these agents. For pre-qualification over the phone, we compared the best AI voice agent platforms.

This comparison is rechecked every quarter, including prices, whether the tools still exist, review scores and screenshots, and the regulatory timeline with them. The last check was September 2026.