Your Agent

Memory

Your agent builds a persistent memory of who you are, what you care about, and how you work. This memory grows over time and follows you across every device.

How memory works

Transistor maintains three distinct types of memory for your agent. Each type serves a different purpose and is stored separately.

Type	What it stores	Example
Episodic	Specific things that happened	You mentioned preferring async communication on March 12
Semantic	Facts and preferences about you	You work in product management at a Series B company
Procedural	How you like things done	When summarizing, lead with the decision, not the context

Memory is local first

All memory is stored on your device by default. It is not sent to any AI provider as raw data. When your agent makes a request to Claude or ChatGPT, relevant memory is included as context in that specific request. The provider sees the context for that conversation, not your entire memory store.

Sync across devices

When you enable sync, your memory is encrypted end-to-end and synced through your Transistor account. Only your devices can decrypt it. Transistor cannot read your memory contents.

Editing and clearing memory

You can view, edit, and delete any memory entry from the Memory section in Settings. You can also ask your agent directly: "Forget that I mentioned X" or "What do you remember about my work schedule."

← PreviousWhat is Transistor Next →Skills