EXPLORINEER
en

KR — ZDNet Korea

Who Works Better? SK Telecom Employees' AI Agents Go Head to Head

AI1 min read

SK Telecom held an internal contest where employees showed off AI agents they built themselves. The agents are already used in real jobs, such as checking power bills and tracking attack servers behind malicious smartphone apps.

What happened

SK Telecom ran the final round of 'AX Spark,' an internal AI agent contest, and gave awards to the best entries. Five finalists competed on how well they made work faster and helped the whole organization become more productive. The agents were made by employees and are already used in daily work.

Key numbers

  • 173 entries were submitted from August to October.
  • 5 entries reached the final round.
  • About 180,000 telecom equipment units across the country are covered by one agent (PYLON).
  • More than 5,000 work AI agents were built inside the company this year.
  • 20 top entries were picked through monthly preliminary rounds.

Why the company is doing this

SK Telecom is pushing a plan called 'One Person, One Agent.' Employees get internal AI communities, development platforms, and job-specific training so they can build and use AI themselves. An AI agent is a program that can carry out tasks on its own, not just answer questions.

Why it matters

The contest shows AI being used for real work, not just tests. Agents now handle tasks like automation, network management, customer service checks, and security threat analysis. SK Telecom plans to spread the best examples to other teams, so AI use may grow across the whole company. CEO Jung Jae-heon said AI is changing the way people work, not just helping with small tasks.

What to watch

  • Whether the winning agents are applied to other teams across SK Telecom.
  • How far the 'One Person, One Agent' plan expands in the company.
  • Whether the multi-agent platform ATLAS can handle work tasks on its own in practice.

Ad

Add 150 words. Learn 10 a day.

Dicter word books send only the words you learn to your review notes.

AIThe Register

Forget retraining - AI can now tweak another AI's weights

MIT CSAIL researchers found that frontier AI models such as OpenAI's GPT-6 Astra can rewrite another model's weights directly rather than through training, altering its behavior. In one demonstration, Astra changed only 0.016 percent of Qwen3.5-0.8B's weights, improving MATH-500 performance by as much as sixfold and IFEval instruction-following roughly threefold, without gradient descent. The researchers call the approach "Weight-Space Programming" and plan to build a benchmark and harness for it.

AIThe Register

Open models can do 90 percent of agentic work at a twentieth of the cost of a frontier LLM: Gartner

Gartner analysts Mark Petty and Raymond Paquet argue that open-weight models can handle about 90 percent of routine enterprise agentic workflow steps at parity with closed models for a quarter of the price of the cheapest closed model. They say the best-case saving against a frontier model run on every step is up to a twentieth, though current open models are more verbose and retry more often, so the realistic saving is closer to five or six times. The analysts also predict that by 2030, 90 percent of enterprise operational agent workflow steps will run on open-weight models, and they argue that harnesses, which route work between models, are where the value in AI products lies.

AI전자신문

Korean Big Data Society to hold fall academic event on the 13th: "Expanding data science with agentic AI"

The Korean Big Data Society will host its "2026 Fall Academic Event" on the 13th at the academic conference hall of Kookmin University's main building in Seoul, under the theme "Data Science Expanded by Agentic AI." The society plans to examine the new value agentic AI, built on data science, can bring to AI transformation (AX) and to share research on various strategies and directions. Pre-registration is open on the society's website until the 12th, and attendance is free for those who pre-register.

AIVnExpress

Why are Chinese companies giving away AI models for free?

In just the first seven months of 2026, Chinese companies released 178 open-source AI models, and most of them can be downloaded free of charge by users, including for commercial purposes. The article asks why China is "giving away" AI models.