Sea Limited Rolls Out OpenAI Codex Company-Wide, Accelerating AI-Native Development in Asia
The Chief Product Officer of Sea Limited, one of Southeast Asia's largest technology companies, has publicly shared the reasoning behind rolling out OpenAI's autonomous coding agent **Codex** to the entire engineering organization.
Sea is a giant platform company spanning e-commerce (Shopee), gaming (Garena), and digital finance (SeaMoney). With the adoption of Codex, the company is shifting from traditional "writing code" work toward an "AI-native development flow." Its move is drawing attention as a pioneering case of AI agent adoption among Asian tech companies.
Codex is not merely a code completion tool — it is an agentic development tool that autonomously drives tasks from planning through execution. A company-wide rollout at a large organization like Sea suggests that the "practical stage" of AI coding has already begun.
FANUC Partners with Google to Operate Industrial Robots with AI Agents
ITmedia reports that **FANUC**, a leading Japanese industrial robot manufacturer, has teamed up with Google to pursue robot operation powered by AI agent technology.
Conventional industrial robot programming centers on manual work that requires specialized expertise. If AI agents can enable natural-language instructions and autonomous task planning, the barrier to robot adoption will drop significantly. In Japan, where manufacturing labor shortages persist, the fusion of AI and robotics is also an urgent challenge.
The FANUC–Google collaboration stands as a symbolic example of generative AI's influence extending beyond software into physical manufacturing floors.
"AI Brain Fry" Spreads Among White-Collar Workers — FT Reports on the Reality
The Financial Times reports that **cognitive fatigue from AI use ("AI brain fry")** is spreading among white-collar workers.
According to the report, everyday use of AI tools leaves employees feeling psychological strain from information overconsumption, constant checking of AI outputs, and the back-and-forth between human judgment and AI suggestions. It is an ironic outcome: AI adoption, expected to boost productivity, is instead increasing mental load.
The report highlights the need to pay attention not only to user experience (UX) but also to workers' cognitive load when introducing AI.
Big Tech Sustains AI Investment by Sacrificing Cash Flows — Economist Analysis
The Economist has analyzed the current state of major tech companies **continuing to prop up the AI boom at the expense of their own cash flows**.
Hyperscalers such as Microsoft, Google, Meta, and Amazon are sharply increasing their investment in AI infrastructure (data centers, GPUs), but monetization has yet to reach a level that justifies the spending. The Economist points out the risk that this investment rush could bring about a "cash-flow winter" rather than an "AI winter."
At the same time, the analysis notes that each company is trapped in a "prisoner's dilemma"-like situation: stopping investment means losing competitiveness.
LINE's New AI Feature Panned as "Too Intrusive" — How to Remove It Also Becomes a Hot Topic
According to ITmedia, a new AI feature added to LINE's message input field is **drawing user backlash as "too intrusive"**.
Many users say AI feels excessively intrusive during everyday message input, degrading the interaction experience rather than improving it. The article also explains how to turn the feature off, making this a case study in just how difficult it is to design the "visibility" of AI features and the user experience around them.
TabPFN-3 Massively Raises the State of the Art in Tabular Prediction
The technical report for **TabPFN-3**, a foundation model for tabular data, has been released. It supports datasets with 1 million rows and outperforms tuned gradient-boosted tree models.
The model is pre-trained solely on synthetic data and is the first tabular foundation model to introduce test-time compute scaling (thinking). It also demonstrates major practical advances, such as beating AutoGluon 1.5 Extreme at 10x the speed. The foundation-model revolution is progressing in tabular prediction — an "unglamorous but important" area of machine learning.
---
**Sources:**
[Sea's View on the Future of Agentic Software Development with Codex](https://openai.com/index/sea-david-chen) — OpenAI News
[ファナックがGoogleと協業、AIエージェントでロボットを操作へ](https://monoist.itmedia.co.jp/mn/articles/2605/15/news087.html) — ITmedia
[White-collar workers report growing feelings of 'AI brain fry'](https://www.ft.com/content/0ba3bd4f-cc3a-4cad-8a8e-76925da2a711) — Financial Times (via Hacker News)
[Big tech is sacrificing its cashflows to prop up the AI boom](https://www.economist.com/business/2026/05/13/big-tech-is-sacrificing-its-cashflows-to-prop-up-the-ai-boom) — The Economist (via Hacker News)
[「邪魔すぎ」LINE入力欄の新AI機能が不評](https://www.itmedia.co.jp/aiplus/articles/2605/15/news084.html) — ITmedia AI+
[TabPFN-3: Technical Report](https://arxiv.org/abs/2605.13986) — arXiv cs.LG