💻 Claude Code Projects rolls out parallel coding for all subscribers 💻
Anthropic removes the waitlist, granting full access to parallel coding threads.

Anthropic has opened Claude Code Projects to every Pro and Max subscriber, clearing the entire waitlist in one move. A Project is an ongoing conversation where Claude coordinates development work, spinning up each task as its own thread running in parallel — meaning multiple code changes, tests, or refactors can progress simultaneously rather than sequentially. The rollout signals competitive pressure against OpenAI Codex and Google's Jules, both targeting the same developer audience. For paying users, this effectively turns Claude into a persistent coding collaborator that maintains context across sessions, a meaningful capability jump that could accelerate adoption among professional engineering teams looking for always-on AI assistance.

🤖 Microsoft ships Decision-1 for fast AI routing 🤖
A new lightweight model rolls out to handle structured decision tasks at speed.

Microsoft released Decision-1, a purpose-built model now available on Microsoft Foundry and OpenRouter. Unlike general-purpose LLMs, Decision-1 is optimized for routing, classification, prioritization, verification, and workflow control — the behind-the-scenes tasks that determine which model handles which query and whether an output passes quality checks. Microsoft reports top performance on structured decision benchmarks, suggesting it can replace heavier models for orchestration logic. This matters because it cuts inference costs for multi-model pipelines. By offloading decision-layer work to a dedicated lightweight model, developers can keep expensive frontier models focused on generation rather than traffic management, potentially reshaping how agentic AI systems are architected.

⚙️ Atomic Machines exits stealth with $250M and Matter Compiler ⚙️
A six-year stealth startup rolls out AI-native manufacturing that turns code into hardware.

After six years in stealth, Jeff Holden's Atomic Machines has emerged with $250M in funding and a bold pitch: the Matter Compiler, an AI-native manufacturing system that converts software instructions directly into working micro-machines — no retooling required. The company's first commercial product is PrimeSwitch, a power switch designed specifically for AI server farms, targeting the surging infrastructure demands of large-scale AI training and inference. This could reshape hardware supply chains by eliminating traditional fabrication steps. If the technology scales as promised, it opens a path to on-demand physical component manufacturing, reducing lead times from months to days for specialized computing hardware.

INTERESTING TO KNOW

🛡️ Nadella pushes 7 AI safety rules for enterprises 🛡️

Microsoft CEO Satya Nadella published an essay urging businesses to adopt a regulation-minded approach: assume AI is already compromised. He proposes seven safety strategies, including mandatory guardrails, emergency brakes that can pause a model mid-task, and a strict rule that no model should ever verify its own work. The framework pushes enterprises toward layered oversight — a notable stance from the company spending billions on AI deployment, signaling that safety narratives are now a competitive differentiator.

🔀 Copilot's Auto mode routes tasks to the best model 🔀

Microsoft EVP Jacob Andreou shared how the Copilot team rebuilt its product strategy around speed and routing precision. The key change: a new multi-model Auto mode that dynamically assigns each user query to the best-fitting model, while low-value entry points were cut entirely to raise per-user engagement. The integration approach mirrors how Decision-1 handles orchestration — suggesting Microsoft is standardizing intelligent routing across its entire AI product stack.