Topic
Explore English articles about AI Agents.
Google is replacing Gemini's Gems with Skills, which reuse instructions in a common file format. But default tools, Workspace sync and Opal migration come with limits, so existing workflows need to be checked.
OpenAI and Synopsys are co-developing GPT-Synopsys, a chip-design AI. We look at how it would operate EDA tools and iterate on fixes, and what Autopilot and a $1B-plus IP deal with Amazon say about changing design workflows and revenue.
Restate has raised $20 million in a Series A. We look at how it records model responses and tool results for recovery, why external APIs still need idempotency, and what Replit's adoption and BYOC pricing show.
Google has announced Gemini 4 Argon. We compare its 1M-token output limit and business-oriented benchmarks with Astra, Opus, and the latest Sol and Sonnet, and look at its limited-time pricing, per-task costs, and restricted rollout.
DDR4 spot prices rose and NAND wafer prices fell at the end of September, yet Q4 contract prices are forecast to increase. We break down the differences across enterprise SSDs, PC SSDs and wafers, and the supply and purchasing conditions behind them.
ASML CEO Christophe Fouquet expects AI's impact to unfold over 10 to 15 years. Company documents show the gap between expanding current EUV output and adopting High NA, plus supplier limits and AI use in chipmaking with Mistral AI.
OpenAI has announced the Decisions API in limited preview. It uses GPT-6 Luna to pick from set options for classification and next actions. We look at how it differs from Structured Outputs, the conditions behind the reported 150 ms figure, and how it compares with Jev.
OpenAI announced ChatGPT Space, a workspace where people and AI co-edit shared documents called Pages. Based on official materials, we look at how it differs from Projects, what recurring updates require, and how links and summaries change who can see your information.
Reuters reports that Anthropic's IPO prospectus warns of existential risks to humanity. We separate the published simulations, an external-access incident during testing, and the company's safety assessment of specific models to show what the warning does and does not establish.
H Company has released Holo4. We look at its screen-and-code training and execution setup, and calculate costs by separating partial-credit scores from full success rates. We also cover open-weight license terms and the limits of verification.
