HR AI is the most regulated AI you will deploy. NYC AEDT requires an annual independent bias audit. The EU AI Act classifies recruitment AI as high-risk. The UK Equality Act and ICO guidance treat indirect discrimination via biased training data as a real liability, not a hypothetical one. We design for that reality from day one — voice-first onboarding, internal HR knowledge agents, recruitment screening with documented bias controls, manager coaching. Not a feature list with compliance bolted on.
HR AI sits where regulation is moving fastest. Bias audits, AI disclosure, candidate consent, data residency. We build with the regulation at the front of the design.
New starters walk through your company by voice. Policies, systems, role expectations, compliance training, attestations. Live whiteboard for the visual parts. Replaces the boring side of L&D with something a new hire actually engages with. Same stack as our Voice AI practice.
RAG over your handbook, policies, benefits, procedures. Employees ask "what is my parental leave allowance" and get a cited answer in their tone of voice. Stops the HR helpdesk being a repository for FAQ resolution and frees the team for actual people work.
CV screening against job criteria, with full audit trail and bias controls. Documented methodology (NYC AEDT compliant), candidate disclosure of AI involvement (CA SB 1001), demographic-blind scoring where law requires. The agent surfaces shortlists; humans make hire decisions.
1:1 prep, performance feedback drafting, sensitive-conversation rehearsal. Agent gives the manager a private space to think through how to handle a tough conversation, with examples grounded in your company's actual culture and policies.
HR AI is the most regulated AI you will deploy. We design to that reality.
NYC AEDT. If you have employees in New York City and your AI is used in hiring or promotion, you need an annual independent bias audit. We design the screening agent to produce the audit data the law requires, not the bare minimum the agent needs to function.
EU AI Act. Recruitment AI is classified as high-risk. Documentation, conformity assessment, transparency to candidates, human oversight. We design for the EU AI Act's high-risk requirements from day one for any UK or EU build that touches recruitment.
UK Equality Act 2010 and ICO guidance. Indirect discrimination via biased training data is the most common failure mode of recruitment AI. We document the methodology, run bias tests against protected characteristics, and publish the results internally for HR and legal review.
California SB 1001 and similar US states. Candidate-facing AI must disclose itself. We build the disclosure into the candidate experience, not as a footnote.
We have built methodology pages with this discipline before. The AI Job Impact Calculator methodology is the public version of how we approach defensible AI, and the same standards apply to HR AI builds.
We map your processes, your tooling, the regulatory environment for your people. Pick the first agent shape, design the compliance layer, scope the build.
One agent built end to end with full compliance layer. Onboarding voice agent, HR knowledge agent, or recruitment screening agent. Bias and audit documentation produced as part of delivery.
We monitor accuracy, retrain as your policies change, run quarterly bias audits. Updates as regulation evolves (and it will). Model and infrastructure costs at cost.
Voice agents that handle onboarding, training, and HR helpline calls. Same stack as Bottie.
RAG over policies and handbooks. Citation discipline so HR answers are defensible.
Where HR AI meets employment law. The same compliance standards apply across both practices.
We work with HR and people teams across the UK, US, and Australia. Voice-first onboarding, bias-aware screening, internal knowledge agents. Compliance built in.
