NEW Claude Projects Changes Everything (with Opus 5.5)

Summary
Content creator Riley Brown presents an in-depth walkthrough and review of Anthropic’s updated "Claude Projects" feature within the Claude desktop, web, and mobile apps. He demonstrates how the new system functions as an agent orchestrator—allowing a central coordinator chat to dispatch tasks to parallel worker threads that execute actions, generate interactive artifacts, and build design boards.

What is shown

• Architecture overview [00:42 - 03:33]: Demonstrating existing projects ("Site Manager", "Long Form Expert") where a primary coordinator chat delegates specific tasks to independent threads (e.g., creating a Composio skills article with an interactive diagram artifact).
• Creating a project and parallel threads [03:57 - 06:30]: Setting up a new project named "Short Form + Twitter" with an explicit goal statement, then prompting the coordinator to launch two concurrent threads—one researching top Instagram transcripts using web scraping tools and another researching short-form scripting strategy.
• Artifact generation and review [06:31 - 07:05]: Inspecting generated artifacts within the thread view, including scraped post breakdowns and script templates, and demonstrating in-line editing and comment annotations.
• Token usage dashboard [09:59 - 12:06]: Opening the project usage drawer showing 5-hour and weekly plan limits, credit balances, and granular per-thread token metrics (e.g., 87.1M tokens total across 7 threads, cache read/write ratios, and coordinator overhead).
• Embedded Design Mode [12:07 - 15:20]: Generating a multi-screen visual design board in "Japandi" style directly from a thread, followed by selecting UI elements, adding contextual feedback comments, and having Claude revise color palettes and layouts in real time.
• Slide decks and artifacts library [16:35 - 17:20]: Converting research findings into an editable, multi-slide presentation deck complete with fetched brand logos and structured layouts.
• Mobile app integration & voice editing [17:29 - 19:40]: Accessing projects, threads, and slide decks via the iOS Claude app, using mobile voice mode to dictate slide revisions hands-free.
• Coordinator vs. Thread capability matrix [20:20 - 21:13]: Reviewing a comparison table detailing the separation of responsibilities between the main chat (planning, memory, delegation) and worker threads (tool execution, code running, connectors, artifact generation, scheduled routines).

Claims & numbers

• Riley Brown states that he tested the updated Claude Projects feature continuously for 48 hours straight prior to recording [00:17].
• The usage analytics drawer displays a project total of 87.1 million tokens across 7 threads, with the coordinator accounting for 17% (14.8M tokens) and worker threads consuming the remaining 83% (72.6M tokens) [10:55 - 11:17].
• The usage panel shows a cache hit rate of 90% and lists individual thread token consumptions ranging from 1.3M to 28.7M tokens [11:11 - 11:25].
• The presenter notes that high-capability models such as Astra and Fable 5.1 are resource-intensive, making monitoring token limits and switching to models like Opus 5 or Sonnet essential for managing rate limits [10:00 - 10:25, 22:07].

Notable quotes

• "You can think of this version of Claude Code Projects as an organized agent orchestrator." [00:32]
• "What Projects does is it separates your main orchestrator agent from the threads within it." [01:22]
• "The work is done in the threads, the orchestration is done by this chat, and all of it lives within this folder." [21:42]

Assessment
This video is a hands-on workflow demo and feature review of the Claude Projects orchestrator UI across desktop and iOS. The presenter demonstrates live multi-agent execution, token tracking, and mobile voice interaction without simulated cuts, though tasks such as web research and slide compilation are shown after completion.

Described by gemini-3.8-flash on 2026-09-29 from the video's audio and frames.