Claude Projects or Liminary

Choose between conversation-based AI (Claude Projects) or a persistent knowledge system (Liminary).

Choose Liminary if you want your client project context to follow you while you write, and meetings rather than being locked inside a chatbot conversation.

Claude Projects lets you create workspaces with uploaded documents and custom instructions. For many professionals, it's become the go-to tool for client-specific AI work: upload your files, set the context, and have focused conversations. Claude's reasoning quality is excellent.

But Claude Projects has a fundamental limitation: each session is a conversation, not a knowledge base. Your uploaded files sit in the project, but they aren't indexed, structured, or connected to anything outside that conversation. And the moment you switch to a different tool, Google Docs, your browser, a different LLM, the context stays behind.

Where Claude Projects stops

Claude Projects works within Claude. That sounds obvious, but it matters. Your project knowledge doesn't follow you into Google Docs. It doesn't capture your meetings. It doesn't monitor your market. It doesn't surface relevant context while you write a proposal somewhere else.

The upload workflow is also manual and repetitive. Every document must be explicitly added to the project. If you want to work with content from a meeting, a web page, or a saved article, you copy, paste, or upload it by hand. Power users we've interviewed describe maintaining dozens of projects across clients, each with its own foundational documents and instructions. It works, but it's a maintenance job.

And if you decide to switch from Claude to ChatGPT or Gemini for a specific task, your project context doesn't come with you. Every tool is an island.

Where Liminary goes further

Liminary builds a persistent knowledge base that sits underneath whatever AI tool you prefer. Save content from anywhere (Chrome extension), capture meetings (Google Meet), upload documents, and organise everything into collections by client or project. Then query that knowledge from within Liminary, or have it surface automatically while you write in Google Docs.

The knowledge isn't locked inside one AI provider. It persists, it connects, and it follows you wherever you work.

What Liminary adds that Claude Projects doesn't

Chrome extension. Save web pages, articles, and resources to your knowledge base with one click. No manual uploading or copy-pasting into a project.

Recall in Google Docs. Your knowledge surfaces in the sidebar while you write. Claude Projects can only be queried inside Claude's interface.

Google Meet integration. Meeting transcripts flow directly into your collections. In Claude Projects, you would need to manually paste or upload a transcript.

Competitive intelligence agents. Agents monitor topics and companies on a schedule. Claude Projects has no background monitoring capability.

1.4x more accurate on hard questions. In our benchmarking, Liminary scored 77.8% accuracy on hard multi-document questions versus 59.4% for Claude Projects with Opus 4.7 (adaptive thinking). Liminary also responded in 7.3 seconds versus 11.7 seconds for Claude.

Shareable and public collections. Build collections that colleagues or the public can browse and query. Claude Projects sharing requires a Team or Enterprise plan.

Who should use what

Claude is an exceptional reasoning engine. If your primary need is having focused conversations with an AI about specific documents, Claude Projects is excellent.

Liminary is for professionals who need their knowledge to work across tools and across time. If you've built elaborate Claude Projects for each client and wish the knowledge would follow you into Google Docs, capture your meetings, and surface relevant context without you asking for it, Liminary solves that problem.

Many users find the two tools complementary: Claude for deep reasoning and synthesis, Liminary for persistent knowledge management and ambient recall.