How to Set Up Claude's Memory So It Finally Remembers You (Your Second Brain in 3 Layers)
Claude can remember you between chats, but it starts blank. Here's the exact 3-layer setup, a project, a profile document, and memory turned on, that stops you re-explaining yourself every morning.
If you open a new Claude chat every morning and spend the first ten minutes explaining who you are, what you sell, and how you write, you do not have a memory problem. You have a setup problem. Claude can carry context across chats now. It just starts blank, and almost nobody builds the structure that fills it.
Here is the exact setup I use, in three layers. It takes about five minutes once.
Layer 1: a project, not a chat
A one-off chat forgets you the moment you close it. A project is a container that holds knowledge and memory across every chat inside it. Same project, same brain.
Make one project per arena of your work, not one mega-project for everything. Content in one. Client work in another. The book in a third. Mixing arenas is how the brain gets muddy and Claude starts blending contexts that should stay separate. This is the single most common setup mistake.
Layer 2: a profile document in the project knowledge
This is the layer that does the work. One document, dropped into the project's knowledge, that tells Claude who you are and how to behave. The more specific it is, the less you ever repeat yourself.
A profile document covers five things:
- Who you are. Your name, what you do, who you serve, the one outcome that matters right now.
- What you're building. The current offer or project and this month's priority.
- How to talk to you. Voice rules (e.g. direct, lowercase, no fluff), words you never use, and format preferences.
- What to remember. Your offers and prices, your proof, the links you paste constantly.
- How to help. When you ask for X, give you Y not Z. Push back when you want to be challenged.
That's it. Fill the blanks, paste it in once, and every chat in that project starts already knowing you.
Layer 3: turn memory on
The layer most people miss. With memory enabled, Claude carries what you tell it across chats and sharpens over time instead of resetting. You correct it once, not every single session.
Why this beats prompting harder
A good prompt helps one chat. A profile in project memory helps every chat, forever. That is the whole difference between fighting the tool and operating it. You stop being the context, and start owning it, as a file you control rather than a chat you lose.
I packaged the exact template, the fill-in-the-blanks profile document, into a free one-page blueprint. You can grab it below and have your second brain running in the next five minutes.
Get the template free. The Second Brain Blueprint at closermethod.com/frame includes the copy-paste profile document and the full 3-layer setup. Build it in five minutes.
If you want the full stack, the brain doc plus the systems that run on top of it, Build It Once installs all of it. One time, yours forever.
Frequently Asked Questions
Does Claude actually remember previous conversations?
Yes, when you set it up. Claude offers memory across sessions and project-based knowledge, but both start blank and off by default. You have to create a project, add a profile document to its knowledge, and turn memory on. Once you do, Claude carries your context across chats inside that project.
What is a Claude project?
A Claude project is a container that holds shared knowledge and memory for one area of your work. Every chat you start inside the project draws on the same documents and context, so you do not re-explain yourself each time. It is the foundation of a working "second brain" setup.
What should go in a Claude profile document?
Five things: who you are and who you serve, what you're currently building, how you want Claude to talk to you (voice and format rules), what facts to remember (offers, prices, proof, links), and how to help you (what to give you when you ask for something). Specificity is what makes it work.
Why does Claude keep forgetting what I told it?
Almost always because memory is off or you are working in one-off chats instead of a project with a profile document. Without that structure, each chat starts from zero. Building the three layers, project, profile, memory on, fixes it permanently.
How long does it take to set up Claude's memory?
About five minutes once. You create a project, paste in a filled-out profile document, and enable memory. After that, every new chat in that project already knows your voice, your offers, and your rules before you type a word.
Elisabeth Bierschenk Hitz is the founder of The Closer Method, where she builds AI-powered systems for operators and creators. She also writes about why owned files beat another subscription.