Industry Adoption

Notion AI is shifting to a model where users spend monthly-granted credits. Seal Yehoshua, who leads development, says cleaner context makes usage cheaper, and explains practical ways to keep costs down under the new system.

Ricoh and Ricoh Japan have put into operation a technology that generates detailed 3D models with few gaps from 360-degree camera footage, adopting it for Tokyo Electric Power Grid's building maintenance work. It combines proprietary localization with AI to fill in featureless walls and floors, producing a model in about half a day.

Japan Exchange Group and Hitachi have begun applying AI to operational monitoring of financial market trading systems, aiming to ease reliance on individual experts. The article explains how they handled hallucination risk to reach 97% accuracy.

Outsourced AI systems can score 90 in the vendor's tests yet prove unusable in-house. The article asks whether outsourcing is wise at all and outlines what companies must keep control of to avoid failure.

Oracle will be the first cloud outside Google to offer Google's TPUs, available on OCI from 2027. It also announced Nvidia Vera Rubin NVL72 availability, new OpenAI-based products, and a Snowflake partnership.

Nidec announced a 3.2MW in-row coolant distribution unit (CDU). The new unit is set to enter mass production next year.

Sakana AI announced Sakana API Platform for enterprises and opened an early access program. It offers the company's LLMs via API with admin controls to manage members and usage at the organization and workspace level. The API is OpenAI-compatible, so existing apps only need to switch endpoints. Sakana Fugu and Fugu Ultra come first, with Namazu and Fugu Max to follow.

Anthropic filed its IPO prospectus (Form S-1) with the US SEC on Oct 26. It discloses that over 90% of internal code is written by AI and that AI spending exceeds labor costs, and it addresses a wide range of risks AI could pose to humanity.

Autumn brings a run of Japan-focused events by overseas vendors. In that context, the article argues that the more AI is used, the more important physical simulation becomes.

Travel company HIS introduced AI agents and robots to its corporate business-trip quotation work, cutting about 100 working hours per month. It had experimented with AI-based efficiency since around 2023 without reaching the accuracy it wanted. The article looks at what turned that around.

With OpenAI Daybreak, Check Point speeds up the research behind new security protections. Its analysts can turn vulnerability intelligence into customer defenses faster.

Anthropic announced a new GSA agreement offering government partners a 40% discount on Claude Enterprise, Claude for Government and the API across all product surfaces, extended to state, local and tribal governments. A new implementation program launches too; savings are estimated at $1 billion.

Claiming AI saved thousands of hours a month does not by itself cut labor costs. The article asks where the freed-up time actually goes, questioning how hours saved are reported as a result.

Two days after the 2026 Kumamoto earthquake, an IT company built a shelter-status map, only to be told by people on site that it was unusable. The article looks at how the team got past that rejection.

Facing a future labor shortage from population decline, Kyoto Prefectural Government is working to make about 7,000 employees more efficient. It adopted conversational AI across the office, but that alone did not ease the workload on the ground, so it decided not to depend on Copilot only.

Itochu Interactive is launching a corporate AI agent implementation training program using Claude and Claude Code. Rather than teaching general generative AI usage, participants build AI agents around their own work, such as drafting proposals and research reports, aiming to automate real tasks and move from merely using AI to delegating work to it.