Johnny Huynh Johnny Huynh

ChatGPT and Claude Projects: Memory or Organization?

I started using Projects in ChatGPT for two reasons: to keep related chats together and to give ChatGPT background information that could apply across those chats. When I later began creating Projects in Claude, I assumed they would work much the same way. They do not. The difference helped me recognize something about how I use AI tools. I value access to broader context more than I value keeping every conversation neatly organized.

The Difference Between the Two

By default, ChatGPT Projects use what OpenAI calls “default memory.” A project can draw on relevant memory from outside the project, and information from the project may also be available in conversations elsewhere in ChatGPT.

ChatGPT also offers a project-only memory option. With that setting, conversations inside the project can use context from that project, but the information stays within its boundaries.

As of this writing, Claude Projects work more like ChatGPT’s project-only option. Conversations inside a Claude Project do not have access to chats outside it.

That difference seems small until the subject of a project begins to overlap with other parts of your life or work.

Where a Self-Contained Project Works Well

I have a Home IT Project that includes information about the hardware and software I use in my home office. When I start a new chat there, I do not have to explain my setup again. The project also gives me one place to find previous conversations about technical issues.

This works well in both ChatGPT and Claude because Home IT is a fairly defined topic. Most of the useful information can live inside one project. The limitation becomes more noticeable with broader subjects such as travel, personal finance, or journaling.

A vacation project might benefit from information about past trips, hotel preferences, loyalty programs, budgets, or research completed in a separate chat. Those details may not have been created with a particular vacation project in mind, but they can still improve the advice.

With ChatGPT’s default memory, some of that context may be available. For example, I might start researching California beaches in a regular chat on my phone. Weeks later, I could begin planning a California trip inside a Project. ChatGPT may be able to connect the earlier research to the itinerary I am now building.

In Claude, that earlier chat would remain outside the Project unless I had started it there or moved it afterward.

What I Learned About My Own Workflow

Claude Projects offer useful organization, but they require me to be more deliberate about where every conversation belongs. I have to decide where to start a chat before I know where the discussion may lead. If I begin outside the Project, I have to remember to move it later. Any useful context that remains elsewhere is unavailable inside the Project. Over time, I realized that this tradeoff does not fit the way I naturally work.

I rely on AI tools to connect information across different conversations. A question about travel may relate to finances. A technology discussion may connect to a work habit. A journal entry may influence how I think about a decision several weeks later. Those connections matter more to me than having every chat filed under the right heading.

I now use Claude Projects selectively. They work best for subjects that are narrow, repeatable, and mostly self-contained. For broader topics, I usually keep conversations in my main chat history. I use standardized prefixes in the chat titles so I can search for related discussions without placing them behind a project boundary. It is less tidy, but it better supports the way I think.

Organization Should Support Context

Projects are more than folders. Their memory settings affect what the AI can bring into a conversation. A well-organized Project can still feel limiting when useful context sits somewhere else. A less structured chat history may be more valuable when the assistant can connect information across it.

The best setup depends on what matters most to you.

For me, continuity wins.

Read More