The 15 AI recruiting tools to know in 2026

The phrase "AI recruiting tool" does not describe one product but a family of products that do very different jobs. A job ad generator, a sourcing engine, a resume scoring tool, an interview note-taker, a skills test and a scheduling agent do not share a price range, a legal risk profile or a benefit. Putting them in a single ranking makes no sense: this page sorts them by hiring stage instead.

Fifteen tools are documented below, two or three per stage, with a bias towards those actually used in France. None of them covers all six stages, and the vendors that come close do it by stacking modules billed separately. Prices collected on 17 September 2026 from the vendors’ own websites and subject to change. When a vendor does not publish a piece of information — an amount, a server location, a trial length — the matching cell reads "not disclosed" rather than an approximation.

This is the stage where general-purpose generative AI is the most useful, and by far the cheapest. Writing a job ad, adapting it to three channels, turning an internal job description into readable copy, preparing an interview outline, translating a posting: none of that needs a specialised tool. Three assistants dominate professional use in France, and all three have a permanently free tier that is enough to form an opinion.

ChatGPT, published by OpenAI, is the most common entry point: a limited free tier, then ChatGPT Plus at $20 per month for individual use, and Business plans billed per seat for teams. The interface is available in French. Data location is not stated on the public pricing pages: that is the point to check before pasting in anything containing a candidate’s name.

Claude, published by Anthropic, offers a free tier, a Pro plan at $17 per month billed annually ($20 monthly) and a Team plan at $20 per seat per month billed annually. Mistral Le Chat, published by the French company Mistral AI, shows a free tier, a Pro plan at €14.99 excl. VAT per month and a Team plan at €24.99 excl. VAT per user per month. Mistral advertises data residency options for Enterprise deployments, but the consumer pricing page does not state the default location.

Sourcing is the stage where the phrase "AI recruiting agent" gets used the most: automating profile search, outreach message writing and follow-ups. It is also the most expensive stage in the table, because the cost of candidate data sits on top of the cost of the software.

LinkedIn Recruiter remains the reference tool for structured talent acquisition teams, and now ships an AI assistant that takes over part of the search and screening work. The interface is available in French. No price is published: the page states that pricing depends on company size and the number of Recruiter seats, and points to a sales consultant. Neither data location nor a free trial is mentioned.

Leonar, a European vendor, targets agencies, executive search firms and internal teams running several mandates at once. Its grid is public: €79 per user per month for the Starter plan prepaid annually (€948 per year), €119 for the Professional plan which adds multi-channel outreach and AI sourcing, with a 7-day free trial. Jarvi, a French vendor, bundles ATS, CRM, multi-channel sourcing, note-taking and an AI agent into a single subscription at €133 per user per month billed annually (€1,599 per year), with a free trial of undisclosed length. Neither states where its servers are on its public pages.

This is the stage where AI replaces the most human time: reading one hundred to five hundred applications and producing a defensible shortlist. Three very different approaches coexist — a scoring tool plugged into what you already run, an ATS with a built-in rating layer, and a technical parsing component sold to software vendors and large HR departments.

ResumeRank is a resume scoring and analysis tool, not an ATS: it reads a resume in about thirty seconds, compares it to your job description and returns a score out of 100 with a written justification, a reading of the skills and an exportable PDF report. Data is hosted and processed in France, in Paris. The Pro plan is listed at €29 per month (€24 per month billed annually) for 200 monthly analyses, and the Enterprise plan at €89 per month (€72 per month billed annually) for 500 analyses shared across a team, with API access and ATS integrations. The interface is available in French and a free trial is offered.

Manatal is an international ATS whose AI engine scores profiles against the requirements of the role: $15 per user per month billed annually ($19 monthly) for the Professional plan, with a 14-day free trial. The vendor advertises a French interface; data location is not stated. Textkernel, finally, does not address recruiters but HR software vendors, job boards and large HR departments: it is a resume parsing and normalisation layer, billed from $99 per month in credits, with 500 free credits to start. French-language resume parsing is supported and the vendor advertises a choice of hosting region between the United States, the European Union and Asia-Pacific, without naming the country.

Interviews are the least equipped stage in French HR teams, and the one where the payoff is most immediate: transcribe the conversation, turn it into a structured write-up and fill in the evaluation grid on the recruiter’s behalf. Two vendors cover this need with public pricing.

Noota is a French vendor and its interface is in French. The free tier caps at 300 minutes per month; the Pro plan is listed at €19 per user per month billed annually (€29 monthly), and the Business plan at €39 annually (€49 monthly). A dedicated recruiting offer, considerably more expensive, adds agent-led interviews and candidate reports. The pricing page mentions a data centre located in the European Union, without specifying the country.

Metaview is an English-speaking vendor aimed at recruiting teams that want to automate interview note-taking, application review and reporting at once. Its grid shows a free entry tier, a Pro plan at $100 per user per month and a Max plan at $300. Neither a French interface nor a data location is documented on the public pages, so both cells read "not disclosed".

Assessment is the oldest segment of this market, and the one where AI is most tightly framed: a test that contributes to a hiring decision has to be explainable to the candidate. The three vendors below address very different organisation sizes, which explains an entry-price gap of one to several dozen.

AssessFirst is a French vendor that measures personality, motivation and reasoning, then models expected performance in a given role. The interface is in French and the vendor explicitly states that sensitive information stays within European borders, without naming the country. Subscriptions start at €499 per month, then adjust to actual usage; no free trial is advertised on the pricing page.

Maki People, also French, offers skills-based screening, a voice interview module and a deeper assessment, orchestrated on top of the existing ATS; the customers it cites are large accounts. No amount is published: pricing depends on company size and the number of candidates assessed per year. TestGorilla, an international vendor, offers a library of more than 300 skills tests with a free tier capped at ten credits per month, then a Core plan at $142 per month on annual commitment and a Plus plan from $400. A French interface exists; data location is not stated.

The last stage is the least visible and yet the one that loses the most candidates: answering fast, pre-qualifying and booking a slot. This is the territory of conversational assistants, often presented as AI recruiting agents.

Paradox, with its assistant Olivia, is the reference in this segment. The product handles applications by instant messaging, pre-qualification, self-service interview booking, video interviews, high-volume hiring events and satisfaction surveys. The target is explicit: high-volume employers — retail, restaurants, healthcare, logistics, manufacturing. No price is published, the page pointing to a demo request instead. The vendor advertises detection of more than one hundred languages and automatic message translation, without specifically mentioning a French interface; data location is not stated.

For writing a job ad, preparing questions or rephrasing an email, yes — and the free tier is often enough. For screening resumes at volume, no. That is not a judgement on the quality of the model, but a question of what surrounds it. Three concrete limits come up every time, and they are detailed in our article "Why ChatGPT isn’t enough for professional recruiting", linked at the bottom of this page.

This section is a factual reference point, not legal advice: for a decision that commits your organisation, talk to your counsel or your data protection officer. Two texts overlap in 2026 for any enterprise AI recruiting setup: the GDPR, which governs the processing of candidates’ personal data, and the European regulation on artificial intelligence, known as the AI Act.

The AI Act classifies AI systems intended for recruitment and candidate selection as high-risk uses, which brings reinforced obligations for both the provider of the system and the employer deploying it: documentation, risk management, data quality, logging, human oversight and information of the people concerned. The application timetable is phased, and the exact obligations depend on your role — provider or deployer.

On the GDPR side, three points shape practice. The employer remains the data controller; the vendor is a processor, which requires a contract meeting article 28. Candidates must be informed that an automated tool is used in the process. Finally, a decision producing legal effects cannot rest on entirely automated processing: the score serves to order and to document, while the decision to reject or to invite remains human. The French data protection authority also recommends keeping an unsuccessful candidate’s file no longer than two years after the last contact, unless the candidate agrees otherwise.

None of the fifteen tools on this page is "the best AI recruiting tool of 2026" in absolute terms: they do not address the same stage, the same volumes or the same budgets. The method that works is to write your constraints down, then eliminate. Six criteria settle it in almost every case, in this order.

This comparison is a documentary survey: every row in the table comes from the product and pricing pages published by the vendors themselves, consulted on 17 September 2026. Nothing has been estimated, extrapolated or copied from a third-party directory, and no review, rating or ranking is produced. When a piece of information does not appear on the vendor’s site — a price, a hosting location, a trial length, the availability of a French interface — the cell reads "not disclosed" rather than an approximation.

Prices collected on 17 September 2026 from the vendors’ own websites and subject to change. Amounts often depend on the scope negotiated: number of users, credit volume, optional modules, setup fees, commitment length. Prices shown in dollars are shown that way because the vendor publishes them so. Always confirm the figure with the vendor before deciding. This page is refreshed quarterly. As ResumeRank publishes one of the tools listed, it appears inside its own category, with no rank and no rating, and with an explicit note about what it does not do.