A list of memories tells you what is stored but not how any of it connects. The memory graph shows the whole thing at once.

Opening the graph

Memory lives in your AI settings. It opens as a graph rather than a list:

• Pan and zoom by gesture, the same way you move around a board
• Hover a folder to see what is inside it without opening it
• Open any note to edit it without leaving the view
• Hubs and spokes are weighted, so the hierarchy reads at a glance instead of looking like a uniform cloud of dots
• Save the graph out as a picture

How memory is filed

Every memory note carries two things beyond its text:

• A path, so related notes sit together rather than in one undifferentiated pile
• A record of where it came from, which keeps what you actually said separable from what the AI worked out on its own

That second one is the part worth knowing. When something the AI believes about you is wrong, the graph tells you whether you told it that or it inferred it.

What writes to memory

The AI saves and recalls on its own as you work. But anything that would change how it behaves is proposed for you to accept rather than applied quietly.

Filling it without doing the work

Two shortcuts, covered in more detail in the memory article:

• Nominate boards you have already built, and the AI reads them and distils what it finds
• Import what another assistant already knows about you, with a short guide per provider

Use Cases

• See at a glance what the AI has connected to what, before you correct it
• Check where a particular belief about your work came from
• Prune what is no longer true after a project ends