Case study · AI System
HHA (Haluan HR Agent): Agentic AI for HR operations
- Client
- Haluan Digital Network
- Year
- 2026
- Sector
- Creator Commerce · HR
An AI-powered HRIS built around the idea of '1 approver, 0 operators': agents that run recruitment, onboarding and daily workplace ops over WhatsApp.
Context
Haluan Digital Network is a large group with thousands of creators and hundreds of brands. Its HR operations spanned many business units, each with its own requests, approvals and manual follow-ups.
The challenge
HR requests were fragmented across chats, forms and manual operator workflows. Approvals stalled, data was duplicated, and every routine task still needed a human operator in the loop.
- Manual approval bottlenecks
- Data duplication across units
- No single source of HR truth
- Operational cost of routine work
- Poor transparency for staff
My role
- Product strategy
- System architecture
- AI agent design
- Full-stack development
- Automation (n8n)
- LLM integration
Approach
How it came together.
- 01
Designed around '1 approver, 0 operators': agents do the work, a human only approves.
- 02
Scoped to the highest-friction flows first: recruitment, onboarding, leave, assets, procurement, payslips.
- 03
Chose WhatsApp as the interface so staff needed no new app to learn.
- 04
Documented the system in a 21-part technical spec before building.
Architecture
Claude-powered agents orchestrate HR flows through n8n, backed by Supabase as the system of record, with WhatsApp as the conversational front end and Next.js for the admin layer.
What changed
- Routine HR tasks handled by agents end-to-end
- Approvals centralized to a single step
- A reusable pattern for AI ops across the group
The hardest part wasn't the AI. It was mapping messy, human HR processes into flows clean enough for an agent to run reliably.
