
Claude Code Projects: One Conversation Coordinates Parallel Cloud Threads
Claude Code Projects is a coordinator chat that splits work into threads, runs them as parallel cloud sessions, and keeps going after you log off.

Expert profile
Founder & AI Product Leader
Yash Thakker is a Generative AI expert with over 12 years of experience in product leadership and technical strategy. As the founder of explainx.ai, he has taught over 300,000 learners and built AI platforms serving millions of users globally. He specializes in Agentic AI, Multimodal RAG, and the intersection of LLMs with consumer hardware.

Claude Code Projects is a coordinator chat that splits work into threads, runs them as parallel cloud sessions, and keeps going after you log off.

Claude Team orgs can now publish skills and plugins org-wide — the default policy is Open (no review step) unless an admin changes it.

Exa launched Snapshot on Sep 18, 2026 — search 400B+ historical webpage versions by date, built for eval leakage prevention and backtesting.

Figure's Helix 2.5 tidied, folded towels, and made beds in 30 unseen homes with zero fine-tuning. The Index-pretraining ablation shows why it matters.

GitHub migrated Copilot's agent runtime from TypeScript to 832K lines of Rust in 14.5 weeks — one engineer with Copilot doing 61% of tool calls.

Google shipped antigravity-preview-09-2026 for Gemini managed agents, plus a Files API and a Credentials API that keep secrets out of the sandbox.

Google Labs launched CC on Sep 18, 2026 — a family AI agent that syncs calendars, drafts meal plans, and fills out paperwork. US-only, 18+.

Google Research launched generative UI with learning-design guardrails — teachers can generate interactive STEM simulations, backed by 30+ samples.

Two builders report GPT-6 Astra solving unsolved WWII Enigma and WWI ciphers — one verified by a cryptologist, one with an open question.

IFM released Uno on Sep 17, 2026 — a diffusion adapter for autoregressive LLMs delivering up to 2.2x faster generation via a lossless sampler.

A Sep 16, 2026 Cambridge paper proposes writing run-time context into an LLM's weights via a hypernetwork instead of the prompt, freeing context.

Meta shipped Muse for Mac on Sep 18, 2026 — it works across apps, files, calendar, and messages. Here's what's new and the open security questions.