Claude Projects

Claude Projects is a feature of Claude, the chat assistant from Anthropic, that groups related conversations into a dedicated workspace with its own knowledge base and custom instructions. Anthropic launched Projects on June 25, 2024 for Claude Pro and Team subscribers, giving each project a 200,000 token context window (about a 500-page book) so that every chat opened inside it shares the same uploaded documents, files, and guidance instead of starting from a blank context. ^[1][2][3] A project keeps documents, files, and chats together, letting a user or a team ground Claude in a fixed body of reference material across many conversations. ^[1][2]

What are Claude Projects?

A project is a self-contained workspace inside Claude.ai with its own chat history, knowledge base, and instructions. ^[6] Before Projects, a Claude.ai user who wanted the model to draw on a particular set of reference material had to paste or re-upload that material into each new conversation; there was no persistent place to store instructions or files that several chats could draw on. Projects addressed this by giving users a container that holds a knowledge base and a set of instructions and then applies both to any chat opened within it. ^[1][2]

Anthropic positions Projects as a way to organize work around a fixed set of knowledge and instructions, distinct from one-off chats. The company says Projects "allow you to ground Claude's outputs in your internal knowledge, be it style guides, codebases, interview transcripts, or past work," and that "Projects can enable idea generation, more strategic decision-making, and exceptional results." ^[1]

When were Claude Projects launched?

Anthropic announced Projects on June 25, 2024, initially for Claude Pro and Team subscribers on Claude.ai. ^[1][4] The Team plan, a collaborative subscription tier, had launched the previous month, and Projects was one of the first features to take advantage of its shared workspace. ^[5]

The feature arrived during a period when Anthropic was adding interface features to Claude.ai beyond plain text chat. Five days before Projects, on June 20, 2024, the company released Claude 3.5 Sonnet along with Artifacts, a side panel that displays generated content such as code, documents, and diagrams next to the conversation. ^[3][4] Projects could be powered by Claude 3.5 Sonnet and was presented as a companion to Artifacts rather than a replacement for it. ^[1][3]

What plans include Claude Projects?

Availability changed over time. When Anthropic introduced the Claude Enterprise plan on September 4, 2024, Projects and Artifacts were included as part of that tier as well, with Enterprise offering a larger 500,000 token context window than the standard 200,000. ^[5] Anthropic later extended Projects to all Claude.ai users, including those on free accounts, with free users limited to creating a maximum of five projects and the more advanced retrieval and sharing capabilities reserved for paid plans. ^[2][6]

Scroll sideways for more →

(Sources: ^[2][6])

How do Claude Projects work?

Inside a project, a user can upload documents, code, and other files that Claude treats as background for every conversation in that workspace. The same set of files does not have to be re-supplied for each new chat. ^[1][2]

Each project can also carry custom instructions that shape how Claude responds across all of its conversations. Anthropic says these instructions let users "further tailor Claude's responses, including instructing Claude to use a more formal tone or answer questions from the perspective of a specific role or industry." ^[1] The instructions and the knowledge base together let a user ground Claude's output in their own material, whether that is a style guide, a codebase, interview transcripts, or past work. ^[1]

What is the Claude Projects context window?

Anthropic described each project at launch as including a 200,000 token context window, which it compared to the length of a 500-page book, so that users could add the relevant documents, code, and notes for a given body of work. ^[1][3] The Enterprise plan raised this to a 500,000 token window, which Anthropic described as equivalent to hundreds of sales transcripts, dozens of 100-page documents, or a medium-sized codebase. ^[5]

For larger knowledge bases, paid plans use retrieval augmented generation. When the material in a project approaches the context window limit, Claude automatically switches to a retrieval mode that loads only the relevant portions of the knowledge base into context for a given question rather than the entire set of files. Anthropic states that Claude "seamlessly enables RAG mode to expand capacity by up to 10x while maintaining response quality," and notes that this enhanced project knowledge is available only on paid plans (Pro, Max, Team, and Enterprise). ^[2][6]

How do teams collaborate with Claude Projects?

On the Team plan, Projects added collaborative features tied to the shared workspace. Team members can "share snapshots of their best conversations with Claude into your team's shared project activity feed," which lets teammates see different ways colleagues are working with the model. ^[1][4] Anthropic positioned this as a way for an organization to spread effective prompting patterns and to bring shared knowledge into common workflows such as product development and research. ^[1]

Sharing of projects themselves is limited to Team and Enterprise plans. Members of those organizations can share a project with individuals, in bulk, or across the whole organization, using two permission levels: a "Can use" level that allows viewing and chatting only, and a "Can edit" level that allows changing the instructions and knowledge and managing members. ^[2][6] Anthropic also stated that "any data or chats shared within Projects will not be used to train our generative models without a user's explicit consent." ^[1]

How do Projects relate to Artifacts and memory?

Projects and Artifacts are separate features that are often used together. Artifacts is the side panel where Claude places generated content (code snippets, text documents, graphics, diagrams, and website designs) so the user can view, edit, and iterate on it next to the conversation. ^[1][3] Projects supplies the persistent context and instructions for the conversation, while Artifacts handles the output produced within it. A chat opened inside a project can still generate Artifacts in the usual way. ^[1]

Anthropic later connected Projects to its memory features. For Team and Enterprise subscribers, Claude keeps separate memory summaries for each individual project, alongside a summary covering a user's non-project chats, so that what Claude remembers in one project does not bleed into another. ^[6]

What changed after launch?

After launch, Projects moved from a Pro and Team feature to one available across Claude.ai, including free accounts with the five-project cap noted above. ^[2][6] The retrieval mode for large knowledge bases and the per-project memory summaries were added for paid users, and the feature was bundled into the Enterprise plan with its larger context window. ^[5][6] Anthropic has continued to describe Projects as a way to organize work around a fixed set of knowledge and instructions, distinct from one-off chats and from its later collaborative and agentic products. ^[2]

References

1. Anthropic, "Collaborate with Claude on Projects." https://www.anthropic.com/news/projects ↩
2. Anthropic / Claude Help Center, "What are projects?" https://support.claude.com/en/articles/9517075-what-are-projects ↩
3. Anthropic, "Introducing Claude 3.5 Sonnet." https://www.anthropic.com/news/claude-3-5-sonnet ↩
4. IT Pro, "Anthropic's new 'Projects' feature looks to supercharge team collaboration with Claude AI." https://www.itpro.com/technology/artificial-intelligence/anthropics-new-projects-feature-looks-to-supercharge-team-collaboration-with-claude-ai ↩
5. TechCrunch, "Anthropic launches Claude Enterprise plan to compete with OpenAI." https://techcrunch.com/2024/09/04/anthropic-launches-claude-enterprise-plan-to-compete-with-openai/ ↩
6. Anthropic / Claude Help Center, "How can I create and manage projects?" https://support.claude.com/en/articles/9519177-how-can-i-create-and-manage-projects ↩