Large Language Models

Today's Stories

1. 1

   LINE Yahoo expands ad delivery with AI agent 'Agent i'

   LINE Yahoo is expanding ad delivery using its AI agent 'Agent i'. The news was reported by日本ネット経済新聞, though specific details about the expansion were not provided in the article body. This move signals a shift toward AI-driven advertising in Japan's digital market. However, the article provides limited context on the implications, so the full impact remains unclear.

   Observers will likely monitor how LINE Yahoo's AI agent integration affects ad targeting and user engagement. No specific metrics or timelines were disclosed.

2. 2

   Furukawa Electric Leads CPO External Laser Source Market

   Furukawa Electric is a top supplier of external laser sources (ELS) for AI data center CPO switches, with high-power InP DFB lasers delivering over 100mW per wavelength at 55°C. The company leverages hard-to-copy physical strengths—MOCVD crystal growth for high-temperature operation and sub-micron precision packaging—to dominate a niche that competitors struggle to enter, aligning with TSMC's COUPE platform and NVIDIA's switch roadmaps.

   Furukawa focuses on choke-point components, avoiding low-margin transceiver competition, and is expanding domestic production in Chiba to meet North American demand for supply chain diversification.

3. 3

   Meta Ends Token-Based AI Evaluations

   Meta formally ended an incentive program where employee performance evaluations depended on how much they used AI tools. New guidance this week replaces references to criteria like 'usage of AI' and the 'AI Native' designation with looser wording, restoring focus on grading employees on their impact. Some Meta employees say the changes free them from feeling pressured to use AI in situations where it doesn't make sense. The shift follows months of 'tokenmaxxing,' where colleagues prompted AI tools repeatedly to maximize internal usage measurements, and a separate project tracking keystrokes that eroded trust. The change may help boost morale and rebuild trust among the workforce.

   Meta is encouraging—but not requiring—employees to test a new agentic AI tool called Hatch, which can browse the web and operate other applications. Some employees fear that if Hatch boosts productivity as intended, Meta could carry out additional job cuts, although CEO Mark Zuckerberg has said further mass layoffs are not expected this year.

4. 4

   Home Depot expands Magic Apron to all U.S. stores

   The Home Depot expanded its AI shopping assistant, Magic Apron, to include localized store knowledge. The tool is now available in all of the company's over 2,000 U.S. locations. Customers can access it via Store Mode in the mobile app or by scanning QR codes on signs. Magic Apron helps in-store shoppers find items, navigate aisles, check availability, and get project advice. It answers questions in the user's preferred language and provides personalized guidance, making the physical store experience more connected. The company says the assistant responds to millions of questions every month.

   Magic Apron builds on the retailer's Store Mode feature launched in 2022. The company debuted Magic Apron in March 2025 and expanded its partnership with Google Cloud in January to develop more capabilities, such as plain-language conversational advice.

5. 5

   Qualcomm debuts enterprise AI PC with HUMAIN

   Qualcomm launched the Horizon Ultra AI PC with HUMAIN, powered by its Snapdragon X2 Elite platform. The device runs AI models locally on the PC, while keeping access to cloud resources when needed. Qualcomm positions this as a response to enterprise demand for on-device AI that improves responsiveness, privacy, and data sovereignty.

   Qualcomm shares gained 7.3% over the past year, versus the industry's 49.5% growth. NVIDIA and AMD are also pushing into AI PCs with competing platforms.

6. 6

   Oracle AI Update: Masked Language Modeling Gains Ground

   Oracle's September 2026 AI update highlights work on masked language modeling for natural language processing, along with new developments in its AI-powered database features. For businesses using Oracle tools, these advances point to more accurate text analysis and smarter data queries—areas where Oracle is competing with AI-native rivals.

   Oracle says the new NLP techniques will help enterprises extract better insights from unstructured data, a key step for automating document-heavy workflows.