AI strategy
AI for HR: what it can do in your company and where the limits are
AI for HR saves hours on job ads, onboarding and staff questions. It is also the area where the law sets the most conditions, because any system that filters or assesses people falls into the high-risk category of the EU AI Act.
Where AI for HR actually helps
When people talk about AI for HR, most picture software that reads CVs and picks the best candidate. That is precisely the use with the heaviest legal obligations and the most room for error. Most of the time AI can save an HR team is elsewhere: in writing, logistics and repeated questions.
An HR manager at a small company spends much of the week writing job ads, juggling interview diaries, preparing paperwork for each new starter and answering the same questions about holidays, payslips or the staff handbook. These are tasks where an AI assistant or an automation works well, and where a mistake is spotted and fixed before it affects anyone.
The boundary lies at decisions about specific people. Who gets an interview, who is promoted, who gets the difficult shifts. AI can help organise information there, but the decision and its justification must stay human, and the law requires this in more and more detail.
Five uses of AI in HR that do not decide about people
Good places to start. All of them handle personal data at some point, so the GDPR still applies.
Job ads and role descriptions
A first draft based on the role's real duties, then reviewed to remove unnecessary requirements and wording that excludes candidates for no reason.
AI for recruiting logistics
Acknowledging each application, offering interview slots and sending reminders. An n8n automation handles this without the AI assessing anyone.
Onboarding
A task list per role, requests for accounts and equipment, a welcome guide and reminders to each manager over the first weeks.
An HR chatbot for internal questions
Answers questions on leave, absence or internal policies from your own documents, cites the source and passes on what it does not know. It must make clear that it is an AI.
Training
Materials, exercises and quizzes tailored to each role, prepared faster and reviewed by someone who knows the job.
The "why not me?" test
Before adding AI to any step of your people processes, imagine a candidate or employee asking why they were not chosen, not promoted or given that shift. If the honest answer depends on what an AI tool said, that use is in the sensitive zone.
Walk through the whole process this way: opening the vacancy, the ad, receiving applications, screening, interviews, the offer, onboarding, day-to-day admin, training, appraisal and exit. Mark each step where AI prepares information or handles logistics, and each step where its output weighs on a decision about a person. Start with the first kind. The second kind needs legal analysis, an impact assessment and real human oversight before it goes live.
AI in HR and the EU AI Act
Regulation (EU) 2024/1689, the AI Act, lists in its Annex III as high-risk the systems intended for recruiting or selecting people, in particular to place targeted job ads, analyse and filter applications and evaluate candidates. It also lists those used to make decisions on terms of work, promotion or termination, to allocate tasks based on behaviour or personal traits, or to monitor and evaluate performance.
The dates have moved. An amendment to the Act, published in the EU Official Journal on 24 July 2026, pushed back the obligations for these systems to 2 December 2027. The ban on emotion recognition systems in the workplace, except for medical or safety reasons, has applied since 2 February 2025.
For a company using a high-risk system, Article 26 requires it to follow the provider's instructions, give oversight to people with the training and authority to exercise it, and inform workers' representatives and affected workers before putting it into use. Watch for a common case: if you use a general-purpose assistant to score or reject CVs, Article 25 can make you count as the provider of a high-risk system, with far heavier obligations. This is general information and does not replace legal advice.
GDPR and labour law: what applies today
There is no need to wait for 2027. Article 22 of the GDPR gives people the right not to be subject to a decision based solely on automated processing that significantly affects them, and rejecting an application does. Processing candidate and employee data with AI also usually calls for a data protection impact assessment, and always for clear information to the people concerned.
National labour law adds its own rules. In Spain, the Workers' Statute gives the works council the right to be informed of the parameters and rules behind algorithms or AI systems that affect working conditions and access to employment. In France, candidates must be told in advance about the recruitment methods and techniques used on them, and the works council (CSE) is consulted on the introduction of new technologies. If you employ people in either country, involve staff representatives early.
Example
An electrical contractor in Valencia hires engineers
Situation: an electrical contractor with about sixty staff hires engineers several times a year. The HR manager loses hours to ads, interview diaries and onboarding paperwork, and answers the same questions about rotas and expenses every week.
Steps: she applies the "why not me?" test to the whole process and rules out using AI to score CVs. She uses an assistant to draft job ads, which she reviews herself. With n8n she sets up the acknowledgement and interview-slot emails, plus an onboarding checklist that alerts the stores, IT and the team lead. She prepares an internal chatbot that answers from the staff handbook and company policies, with a notice that it is an AI and an option to write to a person. She briefs staff representatives before launch.
What changes and how it is measured: she compares admin hours per hire before and after, the days to first interview and the number of questions the chatbot answers versus those it passes on. Each month she reviews a sample of chatbot answers and corrects any that are imprecise.
How to start with AI in HR in five steps
- 01
Find out what is used today
Ask the team which AI tools they use and for what. If someone pastes CVs into a personal account, that is the first thing to fix.
- 02
Check the software you already have
Many recruitment and HR systems now include AI features. Check whether any of them ranks or scores candidates or employees, and on what terms.
- 03
Run the test on your process
Walk each step with the "why not me?" question and pick one or two uses that only prepare information or handle logistics.
- 04
Write the rules and talk to staff representatives
Which personal data may go into each tool, who reviews outputs and how candidates and employees are informed.
- 05
Train the HR team
On their own tasks, with the legal limits clearly explained. Obinova delivers this training and builds recruitment and onboarding automations in n8n, documented so your team can maintain them.
Frequently asked questions
Can AI be used for recruiting decisions?
It can assist, under many conditions. A system that filters, scores or evaluates candidates is high-risk under the EU AI Act, with obligations applying from December 2027, and the GDPR already limits decisions based solely on automated processing. For a small company, the prudent path is to use AI for logistics and leave selection to people.
Is it legal to use ChatGPT to screen CVs?
It is risky. Pasting CVs into a personal account raises data protection problems, and using a general assistant to reject candidates brings you close to the obligations of a high-risk system, possibly as its provider. If you want AI support, use it to summarise a CV that a person will read anyway, in a company account and with candidates informed.
What is an HR chatbot?
An assistant that answers staff questions about leave, absence, pay or internal policies from the company's own documents. It works well if it cites the source, admits when it does not know and hands personal cases to a person. It must state clearly that the one answering is an AI.
Do employees have to be told about AI in HR?
In many cases, yes. The GDPR requires candidates and employees to be told how their data is processed. In Spain and France, staff representatives have information or consultation rights over technologies that affect employment. And for high-risk systems, the AI Act requires informing workers before the system is put into use.
How much does it cost to bring AI into HR?
It depends on what you put in place. Training the team and using assistants you already have costs little in tools. An onboarding automation or an internal chatbot needs design, testing and upkeep, and the cost depends on how many tools are connected. At Obinova the quote comes after a first meeting.