Claude Cowork Adds Skill Recording Feature
Anthropic's Claude Cowork now lets users record custom skills through screen recordings and voice-over explanations, available on Pro, Max, and Team plans.
ARTICLES BY THEO ALMEIDA
LATEST NEWS
-
-
Crowdstrike Finds AI Worm Hiding in Blind Spots
Crowdstrike uncovered a worm targeting AI infrastructure, leveraging blind spots in development environments to evade detection.
-
AMD Demonstrates High-Performance Video Inference with ROCm Libraries
AMD's tutorial shows how to build a video inference pipeline using ROCm libraries, with support for RDNA 3.5 GPUs and HEVC decoding.
-
AMD Demonstrates Efficient MiniMax-M3 Inference on Instinct GPUs
AMD showcases performance of MiniMax-M3, a 428 billion parameter model, on Instinct MI355X GPUs using ATOM and ATOMesh, achieving 340–370 tok/s/user in high-interactivity scenarios.
-
X.ai Introduces Grok Add-in for Microsoft Outlook
X.ai launched Grok for Outlook on July 21, 2026, offering email thread summarization and reply drafting for Microsoft 365 users.
-
AMD Optimizes MiniMax-M3 Inference on MI355X GPUs
AMD announced new optimizations for MiniMax-M3 inference on MI355X GPUs, improving serving throughput and reducing token latency.
-
Army Using Ask Sage AI Tool Exhausts Token Supply
The Army burned through its entire annual token allocation for AI tools in just one service, according to internal emails.
-
Hugging Face Releases GLM 5.2 for Cyber Defense Self-Hosting
Hugging Face disclosed an AI-driven attack on July 16, 2026, prompting a guide for self-hosting GLM 5.2 for cyber defense.
-
Arcade Simplifies AI Protocol for Better Scalability
Arcade announced updates to the Model Context Protocol (MCP), easing server management for large-scale AI deployments.
-
AMD Introduces GEAK V3 for GPU Kernel Optimization
AMD's GEAK v3 boosts HIP and Triton benchmarks by 3.02× and 2.22×, respectively, with support for multiple kernel languages and GPU architectures.
-
Tradeshift Uses Amazon Quick to Modernize Analytics
Tradeshift replaced its legacy BI tool with Amazon Quick, achieving up to 30 times faster query response times and a 40% reduction in total cost of ownership.
-
Couchbase Uses Amazon Bedrock for Multi-Model AI Architecture
Couchbase expanded Capella iQ's AI capabilities by integrating Amazon Bedrock, enabling support for multiple foundation models and improving operational resilience.