Live Last read · last published · next check
A robot arm reaches into a bin packed with a hundred mixed parts and pulls out exactly one without disturbing the rest. On video, this looks simple. In practice, it’s one of the hardest unsolved problems in robotics. This is robotic manipulation: the field concerned with how robots physically grasp, move, and interact with objects in the real world. For teams building Physical AI systems,…
Every robot that folds laundry, picks a part off a conveyor belt, or walks through a warehouse aisle is trying to solve the same problem a toddler solves in its first year of life: understanding the world from its own point of view. That is the core idea behind egocentric data: information captured from a first-person perspective rather than by a fixed camera watching from across the room. For…
Ask an AI chatbot what happens if you push a glass off a table, and it can describe the answer in perfect sentences. Ask a robot to actually catch that glass before it shatters, and most AI systems built on language alone will fail. That gap, between describing the physical world and understanding it, is what AI world models are built to close. The term “world model” has moved from academic papers…
Anyone who has spent time around robotics or Physical AI in the last two years has run into the term VLA model. It shows up in NVIDIA keynotes, Physical Intelligence’s funding announcements, robotics papers on arXiv, and, increasingly, in the roadmaps of companies that have nothing to do with robots yet, because the same architectural ideas are starting to appear in autonomous driving stacks,…
Scale AI built its name as the go-to data partner for frontier labs, autonomous vehicle programs, and government AI projects. For years, its workforce and infrastructure made it the default choice. That changed with Meta’s $14 billion investment for a 49% stake in Scale AI and CEO Alexandr Wang’s move to lead Meta’s own AI lab. For competing labs like OpenAI and Google, Scale is no longer a…
Client overview The client is a leading multinational technology company based in China, developing an AI coding agent designed to solve long-horizon software engineering tasks across multiple programming languages and complex codebases Business Challenges Large-scale, high-quality trajectory data is required to ensure stable learning and strong generalization. Tasks must be diverse and…
Client overview Our client is a research group at a leading U.S. technology university. They are developing an AI Agent in the form of a Computer Use Agent (CUA), capable of interacting with and operating across platforms on macOS, Windows, and Linux. At the current stage, the agent can analyze requests, translate them into goals, and execute the corresponding tasks. Business Challenges Requires a…
Client overview The client is a trusted data solutions partner headquartered in the Netherlands. The company specializes in supporting all stages of AI development, from training to evaluation, by delivering high-quality data through services such as data annotation, generation, and collection. Business challenges The client faced difficulties in seeking a partner who could: Provide a large number…
Client overview The US client is implementing a project to collect user behavior data to train an AI model in a robotic pick-and-place environment. Participants use an AR/VR headset (PICO 4 Ultra) and five motion trackers to record object movement and interaction in different environments (indoor, public, industrial). Business challenges High-precision robotics data requirements: Robotic training…
Client overview A US-based AI client collecting realistic web-browsing interaction data to evaluate AI agent performance. The project focuses on validating step-by-step reasoning, action logic, screenshot fidelity, and final answer quality during real-world website navigation. Business challenges Delayed QA feedback loop: QA review initiated 3 weeks post-production, necessitating large-scale…
