the inner world matters.

most of what makes a person who they are never becomes visible.

• the thoughts they never wrote down.
• the ideas they abandoned.
• the questions they kept returning to.
• the reasons behind a decision.
• the moment something finally clicked.
• the version of themselves they were before they changed.

we tend to preserve the things that come out of us.

• the song, not the eleven versions before it.
• the book, not the pages that were thrown away.
• the shipped product, not the argument that nearly killed it.
• the decision, not the weeks of uncertainty that produced it.

but the outcome is only the surface.

the real story is everything that happened inside before it.

for most of history, that inner world was difficult to preserve. thoughts disappeared when the person did. conversations ended. ideas were forgotten. the reasoning behind a decision became a story remembered differently years later.

we could record what people made. we could rarely record how they became the person who made it.

that is changing.

our thinking now happens through machines that can see it as it unfolds. they witness the questions, experiments, contradictions, discarded ideas, decisions, and reasoning behind them. for the first time, the inner half of our work can become part of our memory.

but seeing it is not the same as remembering it. today, that context is scattered across conversations, applications, models, and sessions. when one ends, much of what happened inside it disappears with it.

innernet exists to preserve that continuity.

context should have continuity.

innernet is a persistent context layer between you and the ai tools you think with. through mcp and an open api, the ai tools you use can read from and write to the same memory.

claude. cursor. codex. chatgpt. your own agents. whatever comes next. you connect once. your context follows.

the memory does not belong to the model. it belongs to you.

memory is more than retrieval.

a larger search index can tell you what you said. we want to preserve what your thinking became.

innernet organises context into maps containing the ideas, people, decisions, projects, and knowledge that make up your world. commits preserve change. branches preserve possibility.

and at the centre of it is netti. netti is the intelligence that tends your memory. it watches what accumulates across your work, connects things that belong together, captures what matters, notices what has changed, and turns scattered interactions into lasting context. you do not have to manually maintain your memory.

• a conversation can become a decision.
• a decision can become part of a project.
• a project can develop a history.
• a history can become part of how your ai understands you.

netti turns what happens into what is remembered.

a position has a history.

most systems move toward the latest answer. something new arrives. something old gets overwritten. the record of changing your mind disappears.

but a person’s context is not a single snapshot. the idea you rejected six months ago may become useful again. a decision makes more sense when you can see what came before it. a contradiction can reveal more than a conclusion.

that is why innernet is versioned.

memory should preserve becoming, not just being.

the memory belongs to the person.

your context should not become a side effect of whichever application you happen to use. innernet sits outside the tools.

each model is a guest. each application is a client. netti is the one that carries the continuity between them.

• move from claude to cursor.
• from chatgpt to codex.
• from one project to another.
• from one model to the next.

your memory does not start over. you can connect different tools to it, build on top of it, and take your context with you.

which is why we own no model and resell none. neutrality is easy to claim and hard to hold when your revenue depends on routing tokens through your own thing. ours doesn’t, and it never will.

we are not building another place where your thoughts live. we are building the intelligence that remembers them.

privacy should exist at the level of memory.

when an ai can access your context, the important question is not simply whether the data is encrypted. it is what this particular tool is allowed to know.

innernet gives facts different levels of access: private, self, trusted, work, and public. permissions are enforced when memory is read. not every tool gets to know everything about you.

what we owe you

a memory has a survivorship requirement other software does not. if a notes app shuts down you lose a notes app; if a memory shuts down you lose the record. so the promises that count are structural.

• you can export everything, right now, without asking anyone.
• deletion is real and immediate; backups are purged within thirty days.
• your memory is reachable over an open endpoint by tools we do not control, so it does not depend on us being the only door to it.
• your maps are never sold, and never train anyone’s model.

built for a world where the tools keep changing.

• the models will change.
• the interfaces will change.
• the applications will change.

your context should not have to start over every time they do.

we believe memory will become infrastructure for the ai era in the same way the web became infrastructure for information. innernet is our attempt at building that layer.

a persistent memory for the things you think, build, learn, and become.