Use cases by function3 use cases
HR and recruiting automation that keeps people making the decisions
HR is the function where European rules are strictest, and they should shape the design from the first day. The EU AI Act classifies AI used to screen, rank or evaluate candidates as high-risk; the Digital Omnibus moved those obligations to 2 December 2027, but GDPR Article 22 already limits decisions based solely on automated processing. Build as if the rules apply now: evidence for every assessment, a person making every decision, and logs that show it.
Some things are simply off the table. Emotion recognition in the workplace has been prohibited under the AI Act since February 2025, which rules out the video-interview analysis tools that were fashionable a few years ago. What remains is useful and defensible: extracting and organizing evidence from applications, running onboarding logistics, and finding the people you already know.
The pages here cover candidate screening that recruiters can explain, onboarding that provisions access on approval and revokes it automatically, and redeploying candidates already in a staffing database. In Germany, a works council has a say in any system that can monitor employee performance, so involve it before the build, not after.
HR and recruiting systems I can build
Candidate screening that quotes the CV behind every score, built for the EU AI Act
Scores each application against a rubric the recruiter wrote, quotes the CV line behind every score, catches hidden instructions, and never rejects anyone on its own.
ATS (Greenhouse, Personio or Bullhorn) / Gmail / Slack / The language model / Audit store (Postgres)
Onboarding that provisions accounts only after approval, and answers new starters' questions
Turns a signed contract into an approved access plan, creates the accounts through APIs, answers new starters from the handbook, and reverses it all on the last day.
HRIS (Personio, HiBob or BambooHR) / Google Workspace / Microsoft 365 / Okta or Microsoft Entra ID / Slack
Redeploying contractors and near-miss candidates before paying to source new ones
Finds contractors rolling off and past near-misses who fit a new job order, drafts personal outreach in the recruiter's mailbox, and updates Bullhorn with every reply.
Bullhorn, Vincere or JobAdder / Outlook / WhatsApp Business Platform / SMS gateway / Postgres with pgvector
Design rules for AI that touches people's jobs
Evidence, not scores
A number between one and a hundred invites trust it has not earned. A quoted line from the CV next to each criterion lets a recruiter check the reasoning in seconds.
Recruiters set the knockouts
Hard requirements such as a license or a work permit are defined by people and applied by code. The model never invents a new reason to reject someone.
Blind review where it helps
Names, photos and addresses can be hidden during first review. It costs nothing and removes a well-documented source of bias.
Access is granted by approval
Onboarding plans are drafted automatically; accounts are created only after IT approves, with least privilege. Offboarding revokes everything, which is the step most often missed.
Outcomes are monitored
Where it is lawful to do so, selection rates are compared across groups over time. A screening process should be able to show that it is not quietly discriminating.
Frequently asked questions
Is AI resume screening legal in the EU?
Yes, with conditions. It is a high-risk use under the EU AI Act, with provider and deployer obligations applying from 2 December 2027, and GDPR Article 22 restricts decisions made solely by automated means. A system that organizes evidence for a recruiter who makes the decision, with transparency to candidates and logs, is the defensible design.
Can AI automate employee onboarding?
It can automate most of the logistics: drafting a provisioning plan per role, creating accounts through APIs after approval, tracking the checklist, and answering new starters' questions from the handbook. It should not grant access on its own, and sensitive questions about pay, health or conflicts should always go to a person in HR.
What does the EU AI Act prohibit in HR?
Among the practices prohibited since February 2025 is emotion recognition in the workplace, except for medical or safety reasons. Social scoring and certain biometric categorization are also prohibited. Screening and evaluation of candidates is not prohibited but is classified as high-risk, with specific obligations.
How do staffing agencies use AI in 2026?
Mostly for matching and redeployment: finding candidates already in the database whose assignments are ending or who nearly got a previous role, drafting outreach, and keeping the ATS current. Screening help is common too, but the recruiter still decides who is put forward to a client.
Also browse: Recruitment and staffing · SaaS · Internal tools and knowledge · all use cases