Author: Carter

Collaborative Robotics’ Proxie 2.0 features two-handed manipulation and autonomous task handling. | Image: Collaborative Robotics Collaborative Robotics (Cobot) has launched the next version of its Proxie mobile robot. The upgraded model offers higher payload capacity, batteries that swap themselves, the ability to identify tasks on its own, and an optional two-armed configuration. The company aims to grow its presence in healthcare, logistics, and manufacturing. The Santa Clara, California-based firm debuted Proxie Gen 2 at Automate 2026, presenting it as a mobile manipulator proven in real-world production settings. “For years, companies had to pick between robots that could move around and…

A common misconception about coding agents is that they’re limited to programming tasks alone. In reality, they’re far more versatile and can handle virtually any office-related task, though their effectiveness varies from one task to another. That said, one particularly popular application that’s drawn significant attention is using coding agents to browse the web — tools like Claude Code and OpenAI’s Codex have become standout examples. These agents have grown remarkably skilled at navigating websites, which opens up a wide range of practical applications. Web browsing can, of course, be useful in many different situations, such as fetching information on…

Today, Sakana AI introduced Sakana Fugu, a multi-agent orchestration platform that operates as a single unified model. You submit a request to one endpoint, and Fugu handles everything behind the scenes. For simpler tasks, it responds directly. For more complex challenges, it automatically assembles and manages a team of specialized models. The intricacies of this multi-agent architecture remain completely invisible to your application code. Key Takeaways Fugu provides a full multi-agent system accessible through a single OpenAI-compatible API. Fugu Ultra currently leads on most published coding and reasoning benchmarks. The orchestrator consistently outperforms the individual models it manages. Opt-out controls…

If you follow Bitcoin, you may have come across “Nostr” — an odd, almost unpronounceable term that stands for “Notes and Other Stuff Transmitted by Relay.” This acronym has fascinated Bitcoin tech enthusiasts for years, and for good reason. Its core concept is straightforward: you don’t need to be tied to a major social media platform to communicate with your friends. Created at the onset of the COVID-19 pandemic by Fiatjaff in March 2020, the Nostr protocol now powers an ever-growing number of purpose-built websites. High-profile figures like Edward Snowden and Jack Dorsey are among its most visible users. Unlike…

Eastman Kodak Company has acknowledged a security breach in which an unauthorized third party gained illegal, temporary access to a stored collection of company data. The hacking collective known as ShinyHunters—responsible for previously targeting Canvas, a learning platform operated by Infrastructure—has claimed responsibility for the attack. According to threat intelligence data shared by Malwarebytes, the group asserts it extracted more than 2.2 million records. Reportedly, this dataset contains both personally identifiable information (PII) belonging to customers and confidential internal business data. Kodak’s Response ShinyHunters issued a final ultimatum with a deadline of June 18, 2026, warning that if their demands…

The Images service, developed using Rust and powered by Workers, operates across every server in Cloudflare’s global edge network. For managing client connections, we rely on hyper, an open-source HTTP library designed for Rust.Last year, we launched the Images binding to support custom, automated workflows for handling remote images within Workers. By late 2025, we redesigned the binding to establish a more direct, local link between the Workers runtime and the Images service.Soon after deployment, we began receiving reports that transformation requests from the binding were failing — though only occasionally and primarily with larger images. What made this particularly…

By Marc Kavinsky, Lead Editor at IoT Business News.According to Berg Insight, the worldwide public carsharing fleet is set to expand from 511,000 vehicles at the close of 2025 to 768,000 by 2030, while the number of users is projected to climb to 141.1 million during the same timeframe. This outlook highlights the growing importance of telematics, reservation systems, and fleet management software as essential infrastructure for shared mobility services.While shared mobility is frequently viewed through the lens of consumer transportation trends, its day-to-day operations resemble a distributed IoT enterprise: thousands of vehicles left unattended, diverse access methods, remote user…

Kinisi Robotics’ KR1 will become part of Bear Robotics’ lineup of service robots. | Source: Bear Robotics Bear Robotics has officially signed a definitive agreement to acquire Kinisi Robotics. Once the deal closes, Kinisi will become a part of Bear. Upon completion of the transaction, Kinisi’s KR1 humanoid robot, its engineering team based in Bristol, and its physical AI expertise will be folded into Bear Robotics, finalizing Bear’s end-to-end physical AI robotics platform. The companies anticipate finalizing the deal within the next few days. “Bear was created to deploy robots in real-world environments, and we’ve spent years developing the infrastructure…

xAI shipped a new mode called /goal inside Grok Build, its terminal coding agent. The feature targets long-running, autonomous task execution. You hand the agent a larger implementation task, then step back. Most coding sessions require back-and-forth execution and verification. You prompt, the agent acts, and you verify each step. /goal changes that loop. The agent keeps working until a task is completed and verified. Verification can mean reviewing code, inspecting webpages, or executing scripts. The mode packages that autonomy behind one command and a visible progress checklist. TL;DR /goal runs long, autonomous tasks inside Grok Build’s terminal agent. It…

Welcome to my series on Outlier Detection. This article focuses on handling categorical data. When tackling outlier detection in tabular data, the initial step is usually transforming the dataset so it becomes either fully categorical or fully numeric. While exceptions exist, this conversion is generally essential: most outlier detection algorithms are built to work with data in a single format, and we must adapt our data to match what the detector requires. When the detector requires categorical input, numeric features must be converted into categories, typically through binning. Conversely, when the detector expects numeric input, categorical features need numerical encoding.…