Getting AX-Ready with hunel

The AI Budget Is There, So Why Hasn't HR Changed?

Now that AI transformation (AX) has risen to the top of the corporate agenda, HR organizations are no exception. Gartner (2025) reported that 76% of HR professionals said "if we don't use AI within the next one to two years, our organization's competitiveness will decline." In fact, nearly 80% of companies worldwide have adopted AI in at least one area of work. (McKinsey, 2025)

And yet something strange is happening. A company adopts an AI chatbot, but HR professionals' workload doesn't drop. It connects HR data to AI, but the insights management wants don't come out. Forrester Consulting's AI digital-workplace survey, released in Q1 2026, shows the cause in data. 85% of responding companies said "for AI to succeed, fragmented data and systems must be integrated first," and 45% of companies that experienced AI underperformance pointed to the absence of organizational context data as a primary cause. Gartner warned in the same direction: "60% of AI projects without AI-ready data will be abandoned by 2026."

Josh Bersin stressed the same point in a 2026 report: "What determines AI performance is not the number of agents purchased but the quality of the underlying data structure." Organizations where AI adoption doesn't translate into results have something in common — they adopt the tool first and think about the data structure later. AI is not a tool but an engine that reads data. Without a structure it can read, even the most sophisticated AI ends up staring at a blank screen.

This article explains, for the CHROs and HR strategists of large enterprises and group companies preparing for AX, the three data structures AI needs to actually work in HR and how hunel supports them.

Data Structure 1 | Conversational HR Data — HR Data Accessible by Natural Language

Why this structure is needed

"How many days of annual leave do I have left this year?" "What are the eligibility conditions for parental leave?" For an employee to get an immediate, accurate answer when they ask AI these questions, there is a precondition: internal HR policies, work rules, and benefits guides must be stored inside the system in a structure AI can read.

This is the core of the RAG (retrieval-augmented generation)-based knowledge management referred to in AX literature. To have AI answer based on internal policy data, that policy must first exist in a searchable form. Yet in many large enterprises, HR policies live in PDF files, benefits guides on intranet boards, and work rules in separate folders. There is no data for AI to connect to in the first place.

The data structure you need

• ① Structured HR policy data — work rules, benefits, and application procedures must be stored in a searchable form within the system for AI to reference them.
• ② ESS transaction data linkage — employee self-service data such as remaining leave, payslips, and certificate issuance must be linked directly to AI responses.
• ③ Onboarding/offboarding process automation — new-hire account creation, guide delivery, and leaver processing must be automatically triggered within the system.
• ④ Structured recruitment data — applications, interview records, and job descriptions (JD) must be stored in a form usable for AI matching.

How hunel supports this structure

hunel's ESS (Employee Self-Service) and MSS (Manager Self-Service) modules structure the HR information employees routinely need — leave, payroll, certificates, request/approval — within the system. The onboarding/offboarding features automate the process from new-hire information registration through leaver processing. The recruitment management module stores the recruitment site, online interviews, applications, and screening data inside the system.

Data Structure 2 | Skill-Based People Data — HR Data That Records Actual Competencies, Not Job Titles

Why this structure is needed

The skill-based organization (Skill-Based Organization) is the most powerful theme in HR for 2026. WorldatWork (2025.12) defined that "the real differentiator in the AI era of 2026 is not technology itself but the Human Readiness behind it." For AI to perform internal talent recommendation, internal mobility matching, and skill-gap analysis, employees' actual competencies must first be recorded inside the system.

Yet in most large-enterprise HR systems, "HR data" amounts to grade, position, tenure, and department. What this person can actually do, which projects they have carried out, and which training they have completed are scattered outside the system. AI cannot do meaningful talent matching without this data.

The data structure you need

• ① Competency/career profile data — not only education, career, and certifications but also completed projects and training history must accumulate per employee in the system.
• ② CDP (Career Development Plan) data — employees' career direction and development goals must be recorded in the system and linked to key-talent and successor data.
• ③ Internal posting/mobility history — internal mobility applications and outcome data must accumulate to become the basis on which AI recommends suitable talent.
• ④ A Talent Search-ready structure — HR data must be structured so that "who has these qualifications" can be pulled instantly with a natural-language query.

How hunel supports this structure

hunel's career-development module manages CDP, mentoring, internal postings, key-talent, and successor data within the system. The position, grade, education, career, and qualification fields of the HR management module are customizable, so they can hold each organization's competency-profile structure as is. Talent Search in the EIS (executive dashboard) provides the ability to instantly find talent matching given conditions.

Data Structure 3 | Continuous Performance Data Pipeline — A Performance and Feedback Pipeline That Flows Without Interruption

Why this structure is needed

In an organization that only has once-a-year evaluation data, AI sees only last year's snapshot. For AI to detect at-risk employees early, analyze performance patterns in real time, and send managers coaching-timing alerts, performance and feedback data must flow continuously through the system. This is the premise on which the "continuous feedback data pipeline" and "risk-signal alert system" described in AX literature operate.

Gartner (2025) reported that "29% of AI's productivity impact comes from operating-model adaptation." The core of operating-model adaptation is moving beyond a once-a-year evaluation structure to one in which performance and feedback accumulate continuously. Without this structure, there is no data for AI to predict from.

The data structure you need

• ① Multi-rater/360-degree evaluation data — results of achievement, competency, and multi-rater evaluations must accumulate in the system regularly so AI can learn and analyze performance patterns.
• ② Labor-cost/headcount status data — labor-cost trends and headcount changes by department and grade must be tallied in real time to enable AI-based workforce planning.
• ③ Incentive simulation data — simulation linking evaluation results to compensation must happen inside the system to maintain data consistency.
• ④ EIS dashboard data flow — the headcount status, turnover, and labor-cost trends management needs must be tallied in real time.

How hunel supports this structure

hunel's evaluation-management module runs achievement, competency, 360-degree, and multi-rater evaluations in an integrated way. The incentive simulation feature is designed so that evaluation results and the compensation system are connected within the system. The EIS (executive dashboard) visualizes headcount status, labor-cost trends, and Talent Search in real time, giving management a structure to check HR insights directly without processing spreadsheets.

The Connection Map Between the 3 Data Structures and hunel

AX readiness starts not from AI tools but from the data structure

The connection between the three data structures and hunel's features can be summarized as follows.

To prepare for AX, some organizations adopt new AI tools first, while others examine the data structure first. The former arrive at the conclusion that "it didn't deliver as much as expected," while the latter end up with an HR system in which AI actually works.

hunel is designed to implement the three data structures above inside an enterprise HR system. The fact that more than 70% of the top 10 group companies in Korea's major industries and over 25% of KOSPI 200 companies chose hunel is not simply because it has many features. It is because it precisely implements the complex HR systems of large enterprises while also being able to put in place the data structure required for AI integration.

Integration with the elizax AI assistant connects the data structure built on hunel to the execution stage of AX. Data structure → AI integration → employee experience visualization. This order is how AX transformation in large-enterprise HR actually works.