Tech Giants Roll Out New AI Coding Tools

- Fujitsu's new platform is powered by its proprietary large language model named "Takane" and utilizes multiple AI agents working in collaboration to automate the entire software development lifecycle. In a proof-of-concept, it reduced a task that would have taken three person-months to just four hours, a 100-fold increase in productivity. - The beta for xAI's Grok 4.2 features a multi-agent system where four distinct AI agents—Grok, Harper, Benjamin, and Lucas—work in parallel and cross-check each other's work. This architecture has been shown to reduce hallucinations by 65%. - Google's Gemini 3.1 Pro is positioned as an advanced reasoning engine for complex coding tasks, capable of planning a full software architecture before writing code. It supports a one-million-token context window and can process up to 900 images, 8.4 hours of audio, or one hour of video in a single prompt. - The new Replit "Skills Search" feature allows developers to find and install pre-built capabilities directly within the platform, moving beyond simple code generation to offer reusable, project-specific knowledge. - Fujitsu intends to use its AI-driven platform to update all 67 of its medical and government software products in Japan to comply with new regulations by the end of fiscal year 2026. - The multi-agent approach in Grok 4.2 is designed for "rapid learning," with xAI planning to roll out weekly updates based on user feedback. - Gemini 3.1 Pro has demonstrated superior performance in competitive coding challenges, achieving a LiveCodeBench Pro Elo rating of 2887, significantly higher than GPT-5.2's rating of 2393. - The underlying goal of Google's engineering productivity research, which influences products like the Gemini coding engine, is to move beyond simple metrics and focus on a holistic view that includes code quality, engineer attention, and developer satisfaction.