Own the Memory, Rent the Model
Models keep getting better and keep getting replaced. The one thing worth owning is the memory that moves between them.
Count the number of times the "best" model has changed hands in the last two years. Anthropic, then OpenAI, then Google, then back again — sometimes within the same quarter. The frontier is a leapfrog race, and that is wonderful for anyone using these tools. It is also a warning about where you choose to keep your context.
The model is not your moat
For the companies building them, the model is everything. For you, the person using them, the specific model is the most replaceable part of the stack. Whatever is best today will be matched in a few months and beaten soon after. If your accumulated context — your chat history, your saved threads, your hard-won prompts — lives inside one provider, you are left with a bad choice every time the frontier moves:
- Switch and start from zero. The new model is smarter but knows nothing about you. You re-explain your work, your projects, your preferences, from scratch.
- Stay and fall behind. You keep the memory but give up the better reasoning, because leaving is too expensive.
Neither is a good deal. Both exist only because the memory and the model are welded together.
What you actually accumulate value in
The thing that compounds over time is not the model. It is your context: what you have read and chose to keep, the decisions you have made and why, the documents and conversations that make up how you think. That asset is yours. It should behave like it — portable, inspectable, and never trapped in one vendor's chat log.
Own the memory. Rent the model.
A memory layer that moves between models
Overlay is built as exactly that layer. Everything you read, write, ask, save, and decide becomes an artifact in your Brain — captured once, connected to everything related, and kept private by default. Then it travels with you into every chat:
- Capture once. Clip a page, drop in a file, save a conversation or a Claude artifact. It lands in your Brain.
- Ask anywhere. Open a chat with Claude, Gemini, or GPT. Your memory comes with you regardless of which one you pick.
- Inspect the recall. Every answer shows what it pulled from your Brain, so the memory is a trace you can audit — not a black box.
When a better model ships next month, you do not migrate anything. You just point your Brain at it. The reasoning gets an upgrade; the memory carries over untouched.
Yours, across every model you will ever use
This is the practical edge of a simple idea: extending your mind has to be personal. A personal memory layer is what makes the difference between a brilliant stranger and a second mind that has been paying attention. One brain. Any model. Yours — this year, and after the next three frontier models have come and gone.
Scale your own thinking
Overlay gives every model the one thing it is missing — your memory. Capture what you read, write, ask, and decide, then bring it back into every chat. One brain. Any model. Yours.
Keep reading
Scaling the Mind
Every tool we have ever built extended the body. The large language model is the first that extends the mind — and the reason Overlay exists.
ReadA Brief History of Human Augmentation
From the hand-axe to the transformer: the tools that pushed past the body — and where thinking finally joins the list.
Read