SK Group is preparing a big organizational restructuring in the second half of this year. The standard is artificial intelligence, especially AI agents. It does not mean leaving the existing work as it is and just adding AI as a helper. It means they want to redesign the organization itself. A main target is Excel-based data work done by office workers. According to the article, they are reviewing a plan to hand this kind of input and organizing work entirely to AI agents. From now on, the key idea is to first check whether AI can do the existing work, and then divide people's work again. The article says that SK Telecom may become a key axis of this change. It also says that Chairman Chey Tae-won has viewed AX as a matter of survival and competitiveness. This flow can be read not as an experiment only by SK, but as part of a bigger change where other large companies are also reviewing a similar direction.
원문 보기
The key of this news is not 'bringing in AI' but 'changing the blueprint of work'
If you read this news as just 'a big company uses AI,' it is easy to miss the main point. The more important part in the article is the phrase rebuilding the organization with AI agents as the starting point. This does not mean buying one more piece of software that slightly helps workers from the side. It means newly dividing which work AI handles first, and what people will review and judge.
If old AI tools were closer to a 'draft writer' or a 'search helper,' AI agents are closer to doing many steps in a row when given a goal. For example, you can let them handle one whole flow: collecting data, matching the format, finding outliers, and even making a first draft of a summary report. If you understand this, you can see why the title uses the phrase major organizational overhaul.
There is one point readers should watch here. From now on, company competition is shifting away from 'whether to use AI or not' to which tasks to hand to AI, and where people keep value and responsibility. If you know this standard, you can read not only SK news but also AI news from other companies with much less confusion.
Bringing in a support tool means adding AI on top of existing work, but an agent-based organization means redesigning the workflow itself.
So the unit of change grows from personal productivity to team structure, approval systems, and sharing responsibility.
What is different between work support tools and an AI agent organization?
| Comparison item | Work support tool | |
|---|---|---|
| Control | A person gives instructions every time | |
Organization built on AI agents AI receives a goal and carries out multiple steps in sequence | ||
| Scope of execution | Single tasks like drafting or summarizing | |
| Approval structure | Keep the existing reporting system | |
| Responsibility sharing | Almost all responsibility for the result stays with people | |
| Performance measurement | Better personal productivity for employees | |
| Organizational impact | Level of training for using tools | |
What work do companies hand over to AI, and what work do they leave to people?
| Decision standard | Work that is easy to hand over to AI | Work that is easy to leave to people |
|---|---|---|
| Repetition | Input and collection repeated every day in the same format | Decision-making where the context changes every time |
| Possibility of standardization | Work with clear rules and templates | Work where negotiation and persuasion matter more than rules |
| Possibility of data conversion | Work where documents, numbers, and conversations remain as digital data | Work with big informal relationships and tacit knowledge |
| Error handling | Work where error costs are low and review is easy | Work where one mistake can grow into legal or reputation risk |
| Core skills | Summarizing, classifying, searching, drafting | Final approval, handling exceptions, judging responsibility, empathy |
| Main examples | Organizing meeting notes, report drafts, finding data outliers | Choosing strategy, negotiation, HR evaluation, external communication |
From Excel to ERP, and back to AI: this is how office work tools in Korea have changed
The sentence 'hand Excel work over to AI' sounds big because office culture in Korean companies has run on Excel for a long time. If you look at the flow, you can start to see why this change is more than simple automation.
Step 1: In the 1980s to 1990s, Excel became the basic language of office work
Excel became more than a program for calculating numbers. It became the basic format for budget sheets, performance tables, and reports. As computerization spread in Korean companies, the culture became fixed: 'organizing numbers = opening Excel.'
Step 2: In the 2000s, ERP came in, but Excel did not disappear
ERP means enterprise resource planning, a tool that puts company information like inventory, purchasing, and accounting into one system. But in real work, the official data stayed in the system, while the final collection and report editing still stayed in Excel. This double structure remained.
Step 3: In the 2010s, systems became wider, but the 'last mile' still stayed in human hands
Matching numbers between departments, explaining exceptions, and rearranging things to fit executive report formats were areas that standard systems were not good at. So a lot of office workers' time in big companies went into organizing Excel files and checking numbers.
Step 4: In the 2020s, AI moved beyond input work into summaries and decision support
Now AI is going one step beyond simple copy and calculation automation. It can read many files, find strange values, write draft reports, and even suggest the next action. So it has started to enter the last mile too, where human hands used to be necessary.
Excel automation and AI agents are not the same kind of automation
| Comparison item | Excel-based work | ERP·BI-based work | AI agent-based work |
|---|---|---|---|
| Data source | Scattered across files | Integrated into the official system | Read multiple systems and documents together |
| Exception handling | Depends on the staff member's experience | Handled within standard rules | Possible to design a structure that detects exceptions and passes them to people |
| Report writing | People reprocess it directly | Automatic generation of structured reports | Helps with summary, interpretation, and even draft writing |
| Decision support | Almost none | Focused on a basic dashboard | Can explain outliers, write comparison sentences, and suggest next actions |
| Collaboration method | File sharing and version management | Centered on a shared system | Work logs, approval flow, and question-and-answer are connected |
| Traceability | Weak tracking of edit history | Focused on system logs | Can keep logs of both the execution process and the reasons for decisions |
There is a history of change behind why SK talks about AX as a survival issue
AX means AI transformation, and it is not just about changing a few programs. If you look at SK's history together, you can understand why this change feels like another big shift.
Stage 1: From a textile company to an energy company
Today's SK originally started as Sunkyung Textile. Later, by acquiring the Korea Oil Corporation, later called Yookong, the group moved its center to energy and chemicals. The first key point is that it did not stay in just one industry.
Stage 2: Taking communications and changing the game again
The acquisition of Korea Mobile Telecommunications in 1994 was a major turning point that became the foundation of SK Telecom. As a company centered on manufacturing and oil refining expanded into an information and communications group, the group's growth engine changed completely.
Stage 3: Adding semiconductors through the Hynix acquisition
The acquisition of Hynix in 2012 became an event that firmly shaped SK's image as a company that changes the big game even in a crisis. After that, semiconductors became a core business that raised SK's status, and the internal sense that change is survival also became stronger.
Step 4: Beyond DX to AX, a declaration to rethink the way work itself is done
Recently, SK has been talking about AX, which goes one step beyond digital transformation (DX). If DX was closer to computerizing and improving efficiency, AX is closer to rebuilding processes and organizations so that AI can actually come in as a real worker. That is why Chairman Chey Tae-won is calling this a survival issue.
Why SK Telecom may move first: telecom is a good industry for adding AI agents
| Comparison item | Telecom company | Manufacturing | Retail |
|---|---|---|---|
| Frequency of customer contact | Very high: plans, consultation, cancellation prevention | Relatively low | High: sales and membership management |
| Real-time data | Network and usage pattern data are generated continuously | Focused on equipment and process data | Focused on inventory and sales data |
| Repeated operations tasks | Consultation summaries, recommendations, outage response priorities | Quality inspection, production planning | Product recommendations, demand forecasting |
| Difficulty of applying agents | A pilot is possible relatively quickly | Need to be careful because of on-site safety and equipment connection | System integration by channel is a variable |
| Main strength | Customer data and operating systems are gathered inside one company | Strong in process optimization | Strong in consumer pattern analysis |
It is not only SK moving: the map of AI agent competition among Korea's big companies
| Group | Main approach | Strength | Point to read now |
|---|---|---|---|
| SK | Group-wide integrated AX and use in telecom and manufacturing sites | SK Telecom, SK AX, and fast restructuring at the group level | The most aggressive point is that it connects even to organizational restructuring |
| LG | Exaone-based affiliate role-sharing model | Division of roles among LG AI Research, LG CNS, LG Uplus, and LG Electronics | Its strength is combining its own model with industry-specific services |
| Samsung | Platform, collaboration tool, and B2B service expansion model | A wide device and solution ecosystem | It is better to look at it together with platform scalability, not only organizational adoption |
When AI comes in, will people really decrease: you can see it by looking at Korea's past automation pattern
This is the question many people are most curious about. But if you look at Korea's automation history, what usually appeared first was task breakdown and role redesign, rather than the total number of workers going down right away. If you know this pattern, you can avoid exaggerated interpretations.
Stage 1: Manufacturing automation raised productivity, but changed the employment structure
In the 1980s and 1990s, automation and robotics made factories more efficient. But as production growth did not lead directly to employment growth, people started using the phrase 'jobless growth.'
Stage 2: The spread of ICT put pressure on repetitive work first
Since the 1990s, computerization and digitalization worked in a way that reduced repetitive work done by mid-skill workers. Rather than jobs disappearing overnight, the job structure split so that some tasks became smaller and some tasks became more important.
Stage 3: Even in service industry computerization, reassignment came before layoffs
In banks and the service industry too, counter work and customer response methods changed, but the changes appeared in complex ways, like branch restructuring, more non-regular workers, and changes in job character. This is why it is hard to explain it with automation alone.
Stage 4: In the generative AI era, routine white-collar work is reorganized first
Now, repetitive office tasks like Excel, documents, consultation, and analysis are becoming the first targets of automation. So instead of asking right away about 'how many people will be cut,' it is more accurate to see which tasks disappear and which review and approval roles remain.
So how should we read this news?
If you read this report as one simple line, like 'AI immediately replaces human work,' you miss too much context. A more accurate way to read it is this. Look at which tasks the company is trying to standardize and hand over to AI, and how people are moving toward final approval, exception handling, and responsibility judgment.
The SK case especially shows three things together. First, office culture centered on Excel has finally started to get direct pressure from AI. Second, the race to adopt AI has now moved beyond model performance to the stage of real workplace settlement and organizational redesign. Third, this kind of change will likely start more seriously first in industries like telecommunications, where there is a lot of data and many customer contact points.
If similar news comes out later, check it like this. More than 'what AI does,' look at which work unit it takes on. If the words 'organizational restructuring' come up, check how the approval system and responsibility structure change. If it says 'employment impact,' see whether changes in the job structure come before the total headcount. Even with just these points, you can judge the next news much more clearly.
The first question is not 'How many people does AI replace?' but 'Which tasks does it take on?'
The second question is 'Is it just introducing a tool, or is it organizational redesign?'
The third question is 'Is it a model competition, or a competition for real workplace settlement?'
We will tell you how to live in Korea
Please give lots of love to gltr life