PLAYBOOK · HR & PEOPLE
A private HR assistant for policies, benefits, and onboarding.
Employees deserve fast, accurate answers about leave, benefits, expense rules, and onboarding — without HR teams answering the same questions 100 times a week. This playbook scopes a domain, indexes policy, and ships a private chat.
HR partners are not search engines, but most of their inbound queries are search queries in disguise. "How many vacation days do I get in Year 3?" "What's the parental leave policy in Sweden?" "Can I expense lunch on a client visit?" VDF AI gives employees a private chat that knows the answer — and gives HR partners back their time for real people work.
The problem
HR is a search problem disguised as a Slack problem
Policies live in PDFs, intranet pages, HRIS UIs. Employees ask the same things. HR partners burn hours on lookups, not on people.
The VDF AI approach
One domain, one source, one assistant
An HR Domain in AgentsHub scopes its tools, indexes, and agents. The Living Knowledge graph ties policy versions to effective dates. Employees ask; the assistant answers with citations.
WHY THIS MATTERS NOW
HR is a search problem disguised as a Slack problem
The economic cost of HR partners answering the same policy question for the hundredth time is staggering — and the cost compounds at every reorganisation, every new geography, every benefit redesign. None of that work is what an HR partner is hired for.
VDF AI carves an HR Domain in AgentsHub, indexes policy with effective dates in the Living Knowledge graph, and ships a private chat experience employees can trust. Every answer is cited to the policy version that was effective at the time the question is asked.
WHAT YOU NEED TO START
Prerequisites for a pilot
Content
- Policy library (PDF, Word, Confluence)
- Benefits documents per geography
- Onboarding checklists
- Employee handbook
Systems
- HRIS read endpoint (Workday, SAP, BambooHR)
- PTO / expense system endpoint
- SSO group claims for scoping
- Sensitive-topic escalation policy
People
- One HRBP partner per region
- One HR-IT owner
- One CX writer for tone
- Optional: legal reviewer for jurisdiction
REFERENCE ARCHITECTURE
HR Domain end to end
PTO · expense · HRIS lookup
Intent: hr-question
PLAYBOOK · STEP BY STEP
From policy PDF to private chat
Create the HR Domain
Carve a domain in AgentsHub. Assign HRBP and HR-IT as owners. Define access rules.
Index policies with effective dates
Living Knowledge captures policy versions and validity windows. The assistant always picks the version current at the employee's location.
Wrap HRIS lookups as Custom HTTP tools
Balance lookup, PTO request, expense category — each is a typed tool.
Compose the HR Assistant Agent
System prompt: cite the policy section; never disclose another employee's data; escalate sensitive topics to HRBP.
Roll out in VDF AI Chat
SSO-bound, audit-logged, and energy-tracked from day one.

OUTCOMES
HR partners back to people work
repeat policy questions hitting HR partners.
answers cite the policy version effective at the time.
employee data leaves the company perimeter.
SEEMR REFERENCE
Living Knowledge keeps answers fresh
Policy amendments enter the graph. SEEMR's knowledge-graph mode rebalances retrieval and routing as policies evolve.
FREQUENTLY ASKED QUESTIONS
What teams ask before shipping this playbook
Will the assistant disclose another employee's data?
No. Tool calls and retrieval are scoped to the requesting employee through SSO claims. The agent is structurally unable to retrieve other employees' records.
How does it handle multi-country policies?
Each country is its own scope inside the HR Domain. The Living Knowledge graph captures effective dates and country tags, so the right version is retrieved per employee.
What about sensitive topics like complaints?
Sensitive topics route to an HRBP with a structured handover note. The agent does not attempt to resolve them.
Can the assistant take action?
Yes — for PTO requests, expense category questions, and address changes. Each action is a typed Custom HTTP tool with audit trail.
How fresh is the policy data?
Re-indexing runs on a schedule per source. Major policy changes can be webhooked to trigger immediate re-indexing.
How long to roll out?
Three to four weeks for a single-region pilot; multi-region rollouts run by region after the first.
RELATED PLAYBOOKS
Continue with related VDF AI patterns
GET IN TOUCH
You Have Questions
Tell us what you’re trying to achieve—governed AI Networks, enterprise RAG, deep integrations, or on‑premise deployment. We’ll help you map the right architecture, security posture, and rollout path. If you’re moving beyond AI pilots and need scalable, auditable execution, reach out—our team is ready to help.