Stay on track with the blogger and advance together in the AI era.
DeepSeek introduces a “cyber finger” that lets AI not only recognize objects but also infer their spatial relationships, orientations, and manipulability, turning visual perception into a digital simulation of touch and enabling more realistic interaction in robotics, AR, and assistive technologies.
Researcher Li Bojie’s “Uncompressible Knowledge Probe” uses random, black‑box API queries to gauge how much irreducible knowledge a large language model retains, allowing an indirect estimate of its effective parameter count and prompting a broader debate on model evaluation and transparency.
Maigret is an open‑source Python utility that, with a single username, scans over 3,000 websites to automatically collect public profiles, images, locations and linked accounts, offering a fast, extensible way for OSINT practitioners to build comprehensive digital footprints while emphasizing legal and ethical use.
TradingAgents introduces a multi‑agent LLM framework that transforms AI from a single‑point price predictor into a collaborative trading team, offering roles such as analyst, researcher, trader, and risk manager, with open‑source code, Docker deployment, and over 59,000 GitHub stars.
Running a 400‑billion‑parameter model locally on the iPhone demonstrates a leap in model compression and edge AI, turning the device into a cognitive agent that handles tasks without apps, while Apple’s upcoming iOS 27 visual‑intelligence features and hardware upgrades cement its role as the core AI ‘digital passport’.
CMU’s new ICLR‑2026 paper demonstrates how AI can move beyond static 3D model generation to create interactive scenes by learning both geometry and functional properties, enabling objects like doors and drawers to be manipulated, a step toward usable simulators for robotics and VR.
Microsoft’s FY2026 Q3 report shows AI revenue soaring to $37 billion, a 123% year‑over‑year increase, while overall revenue hits $82.9 billion, driven by rapid growth in Copilot subscriptions, a 40% rise in Azure revenue, and a $627 billion surge in RPO contracts.
The article introduces mattpocock/skills, an open‑source collection of lightweight shell “skills” that structure prompts and shared context to keep AI coding assistants like Claude Code or Codex from misinterpreting requirements, offering quick installation and configuration steps for engineers seeking reliable, controllable AI‑augmented development.
The article presents Karpathy’s four “sins” of LLM coding and shows how a simple CLAUDE.md file implements four guiding principles—thinking before coding, simplicity, surgical edits, and goal‑driven execution—to make Claude Code produce cleaner, more reliable code, with easy installation and broad applicability.
The article examines DeepSeek V4’s open-source trillion-parameter model and Meituan’s use of an entirely domestic compute cluster, arguing that together they demonstrate China’s emerging dual-track strategy of algorithmic openness and home-grown hardware, signaling a clear move toward a de-Americanized AI ecosystem.
