Nvidia's personal AI supercomputer
At GTC, NVIDIA introduced DGX Spark and DGX Station, personal AI supercomputers powered by the Grace Blackwell platform, bringing data-center-level AI performance to desktops. DGX Spark, the world's smallest AI supercomputer, features the GB10 Superchip, delivering 1,000 TOPS of AI compute with fifth-gen Tensor Cores and FP4 support. It enables researchers, developers, and students to prototype and fine-tune AI models locally before deploying them to DGX Cloud or other accelerated infrastructures. DGX Station, built with the GB300 Grace Blackwell Ultra Superchip, offers 784GB of unified memory and an NVIDIA ConnectX-8 SuperNIC with 800Gb/s networking for large-scale training and inferencing.
Both systems integrate with NVIDIA’s AI software ecosystem, including CUDA-X, NVIDIA NIM microservices, and AI Enterprise software, providing optimized AI development. Reservations for DGX Spark are open now, while DGX Station will be available later this year through ASUS, Dell, HP, Lenovo, and other partners.
Google adds interactive workspace to Gemini
Google has introduced Canvas, a new feature for its Gemini AI chatbot, similar to OpenAI’s Canvas for ChatGPT and Anthropic’s Artifacts. Canvas provides an interactive workspace for users to create, refine, and collaborate on writing and coding projects. It allows users to edit specific sections of drafts, adjust tone and formatting, and export content to Google Docs. Additionally, it includes programming tools, enabling users to generate and preview HTML, React, and web app prototypes with iterative updates.
Alongside Canvas, Google is also adding Audio Overview from NotebookLM to Gemini, which creates podcast-style audio summaries of documents and web pages. Users can upload files, generate summaries, and download or share them. Both Canvas and Audio Overview are now available for free to Gemini users worldwide, though Canvas’ coding preview is currently limited to the web, and Audio Overview is only available in English.
Adobe reveals 10 purpose-built AI agents
Adobe has introduced 10 AI-powered agents and an orchestration tool on its Adobe Experience Platform (AEP) to enhance customer engagement, content production, data management, and site optimization. These agents work in the background to proactively provide insights, automate workflows, and improve user experiences.
Notable agents include the Site Optimization Agent, which monitors website traffic, bounce rates, and broken links, and the Product Advisor Agent, which suggests engagement strategies. Additionally, Adobe launched Brand Concierge, a tool that personalizes websites based on a customer’s past interactions, enabling tailored experiences such as customized greetings and relevant content suggestions.
