HR agent
Grounded answers from your own handbook, screening that ranks and evidences rather than rejects, and onboarding that gets chased to completion across every system.
What does an hr agent do?
An HR agent covers three distinct jobs: answering employee policy questions from your own handbook with citations, screening applications against documented job-specific criteria, and driving onboarding checklists to completion across systems. Screening is the sensitive one — it is regulated in several jurisdictions, so the agent scores against written criteria only, records its reasoning, never makes a rejection decision autonomously, and is monitored for disparate impact rather than assumed to be neutral.
Today
Today HR fields the same policy questions repeatedly over email and chat, reads applications against criteria held loosely in memory, and manually tracks onboarding tasks across IT, payroll, facilities and training systems — following up by hand when something stalls. All three are high-volume and low-judgement relative to the work HR is actually needed for.
Why it hurts
Leave entitlement, expense limits, notice periods and benefits questions consume HR time daily. The answers exist in the handbook, but employees ask a person because searching a PDF is worse than asking.
Senior HR time spent on lookups instead of on people issues.
The same application reviewed by two people, or by one person on a busy afternoon, produces different outcomes. Criteria are informal, so nobody can reconstruct why a candidate was advanced or dropped.
Legal exposure and good candidates lost for undocumented reasons.
A new joiner needs accounts, hardware, payroll setup and training assignments across four systems. When one step stalls, nobody notices until the employee reports it in their first week.
A poor first impression at the moment it matters most.
Runtime behaviour
Every trigger, the action it takes, and whether a human stays in the loop. This is the table to send to whoever has to sign off on the agent's decisions.
| # | Trigger | Action | Oversight |
|---|---|---|---|
| 1 | An employee asks a policy question in Slack, email or the HR portal | Retrieve the governing policy passages, scoped to that employee's entitlements, and answer with citations | Autonomous |
| 2 | The handbook does not cover the question | Escalate to the HR team with the question and the passages that were searched | Human approves |
| 3 | An application is submitted for an open role | Extract structured experience and skills, and score against the documented job criteria with evidence per criterion | Autonomous |
| 4 | Screening score is produced | Present a ranked shortlist with per-criterion evidence to the recruiter — no candidate is rejected autonomously | Human approves |
| 5 | A new hire is confirmed | Create the onboarding plan and raise tasks across IT, payroll, facilities and training | Autonomous |
| 6 | An onboarding task is overdue | Chase the owner, then escalate to the hiring manager if it remains unresolved | Autonomous |
How it is built
Answers retrieved from your handbook and policy documents with citations, scoped to what the asker is entitled to see, and refusing when the policy is silent.
Applications scored against documented, job-related criteria with per-criterion evidence — producing a ranked shortlist for a human decision.
Screening output is monitored for disparate impact across groups on an ongoing basis, because consistent application of biased criteria is still bias.
Task creation across every downstream system, with active chasing and escalation so nothing stalls silently.
Results
Of the three jobs an HR agent does, two are straightforward engineering and one is genuinely sensitive. Policy Q&A and onboarding orchestration are ordinary automation problems. Candidate screening is a regulated activity in a growing number of jurisdictions, and it deserves a different posture.
Our default is that the agent never makes a rejection decision. It scores against documented, job-related criteria, attaches the evidence it found for each one, and produces a ranked shortlist. A recruiter decides. Every decision and every override is logged.
That is not a limitation imposed by the technology. It is the design that keeps the system defensible under New York City’s bias-audit rules, the EU AI Act’s high-risk classification for employment screening, and whatever arrives next.
The common argument for automated screening is that it removes human inconsistency. It does. What it does not remove is bias in the criteria — and it applies that bias with perfect uniformity, at volume, which is arguably worse than inconsistent human bias because it is systematic.
So the criteria have to be examined before automation, protected characteristics and their proxies excluded from the inputs, and outcomes monitored for disparate impact continuously rather than audited once at launch. Recruiter override patterns are useful here: if humans consistently overturn the agent on a particular candidate profile, the criteria are wrong.
An HR agent that states a leave entitlement without showing the policy is creating liability. Grounding every answer in a retrieved passage does two things: employees can verify it, and when an answer is wrong, HR fixes the handbook rather than debugging a prompt.
Next Step
A 30-minute technical call with an engineer who has shipped this before — not a sales qualification round. You leave with a feasibility read, a rough shape for the build, and an honest answer about whether it is worth doing at all.