What are Claude Projects?

Claude Projects are Anthropic's way of grouping related conversations together with a shared set of files and custom instructions. You attach documents to a project, write a project-level prompt that tells Claude how to behave in that space, and then every chat inside the project inherits both. It's excellent for scoped work — a book draft, a codebase, a research thread. It's not designed to be a memory of you across your whole life.

How is Konshus different from Claude Projects?

Claude Projects are project-scoped, single-provider, and don't export as a portable persona. Konshus is life-scoped, multi-provider, and exports persona summaries you can paste into Claude, ChatGPT, Gemini, or anywhere else. They solve different problems: Projects organize a body of work inside Claude; Konshus organizes the body of who you are, independent of any one tool.

Can I export a Claude Project?

You can request a full Anthropic data export (Settings → Account → Privacy → Request data export), which includes project files and conversation transcripts. What you can't do is export the project as a structured persona, import it into a different model, or share it across tools. The export is a backup, not a portable identity.

Does Claude Projects remember me across projects?

No. Each project is its own context. Claude doesn't carry what it learned in Project A into Project B, and there's no global 'Claude knows you' layer the way ChatGPT Memory tries to be. If you want continuity across topics, you have to either rebuild that context per project or hold it somewhere outside Claude.

Can Konshus and Claude Projects work together?

Yes — that's the most common setup. Members import their Claude export into Konshus, distill it into atoms, and then paste a Whisper or Briefing into the top of a Claude Project's instructions. The project's narrow focus (book, codebase, research) stays in the project; the broader context of who they are sits in the Konshus vault and gets pasted in whenever it's useful.

What happens to Claude Projects if Anthropic deprecates a model?

The project files and instructions stay; the model behind the project changes. In practice that means a new model can read the same files but may interpret your custom instructions differently, or lose nuance that a previous model handled well. Konshus stores the underlying context (atoms) separately from any model, so model changes don't degrade the underlying record of you.

Do I need Konshus if I already use Claude Projects?

If your AI usage is fully scoped to projects inside Claude — a manuscript, a codebase, a research notebook — Projects alone may be enough. If you also use ChatGPT, journal with AI, want a single coherent memory across topics, or want to back up the version of you that exists across years of conversation, Konshus is the layer underneath. Most Konshus members keep their Claude subscription.

Is Konshus a replacement for Claude?

No. Konshus doesn't replace any model — it's the memory layer that travels with you between models. You still use Claude for Claude work, ChatGPT for ChatGPT work, Gemini for Gemini work. Konshus just makes sure all three sound like they actually know you.

A comparison · ~6 min read

Konshus vs Claude Projects — Two Different Kinds of Memory

Claude Projects are one of the best things Anthropic ships. They quietly solve the "every new conversation starts from zero" problem inside Claude. Konshus solves a different problem — what happens when the conversation needs to live outside Claude, or across every model you use, or longer than any single provider's memory roadmap. Here's how the two compare and how they stack.

The short version

Claude Projects are a container inside Claude. You attach files, write project-level instructions, and every conversation in that project inherits both. Brilliant for scoped work — a manuscript, a codebase, an ongoing research thread.

Konshus is a separate vault that holds the version of you that any model could use. It ingests Claude exports, ChatGPT exports, journals, voice, documents, and social archives, distills them into structured atoms, and exports a portable persona.

Projects = body of work. Konshus = body of self.

Side-by-side

Dimension	Claude Projects	Konshus
Scope	One project at a time	Your whole life — all topics, all years
Where it lives	Anthropic servers, inside Claude	Your vault — encrypted, exportable, deletable
Which models can read it	Claude only	Any — Claude, ChatGPT, Gemini, Grok, local
What's stored	Files + project instructions you write	Atoms (facts with source + confidence) + voice + documents
Cross-project memory	None — each project is isolated	One unified vault across every source
Portable persona export	No	Yes — Whisper, Briefing, Full Mirror, Crisis Handoff
Survives provider change	No — tied to Claude	Yes — the vault is independent
Cost	Included in Claude Pro / Team	From ~$4/month (Spark, billed annually)

What Claude Projects are genuinely good at

When the work is scoped — a book, a codebase, a long research thread — Projects are excellent. Files persist. Custom instructions tune Claude's behavior for that specific space. Conversations inside the project don't have to keep re-explaining context. For project-shaped work, Projects are the right answer.

The limits show up when work doesn't fit in one project, when context needs to follow you outside Claude, or when you want a version of yourself that isn't tied to any one tool's roadmap.

What Konshus does that Claude Projects can't

Crosses providers. A Konshus persona pastes into Claude, ChatGPT, Gemini, or any other model's system prompt. The same context, everywhere you work.

Crosses topics. Projects isolate by design. Konshus unifies by design. The vault holds your writing voice, your working style, your stated identity, and the patterns that show up across years of conversation — all in one place, queryable by you.

Crosses sources. Projects take files. Konshus takes ChatGPT exports, Claude exports, journals, voice memos, documents, and Facebook / Instagram / LinkedIn archive ZIPs. Years of context that were never going to land in a single Claude project become part of the vault.

Outlasts the tool. When Claude changes, the vault doesn't notice. When a model is deprecated, your context is unaffected. When you want to leave a provider entirely, you walk out with your memory in hand.

Short answer

Claude Projects organize a body of work inside Claude. Konshus organizes the body of you across every model. They're not competitors — most Konshus members keep their Claude subscription and paste a Konshus Briefing into the top of any Project that benefits from broader context. Use Projects for scoped work. Use Konshus when continuity, portability, and ownership matter.

A typical stack

You keep your Claude subscription. Projects stay where they are — manuscript, codebase, research thread.
You import your Claude export into Konshus. The vault distills it into atoms.
You optionally add ChatGPT exports, journals, voice, documents, social archives. The vault unifies them.
When a Project would benefit from broader context, you paste a Whisper or Briefing into its instructions.
When you start a fresh thread in Claude (or any other model), you paste the same Briefing. Every assistant acts like it knows you.

The Project handles the scope. Konshus handles the self. They stack cleanly. Related: Konshus vs ChatGPT Memory and best AI memory systems.

Frequently Asked Questions

Projects organize the work. Konshus organizes you.

Konshus ingests your Claude and ChatGPT exports, distills them into a structured vault, and gives you a portable persona that works with every Project, every model, every provider — encrypted, exportable, yours.

Meet Konshus