Entity Extraction

What is entity extraction?

As you chat with an AI, you drop a lot of key info: "my name is Alex", "budget is five thousand", "I'm free Wednesdays", "allergic to peanuts". Entity extraction is the AI automatically spotting those key entities — names, dates, numbers, preferences, facts — and storing them for later.

What problem does it solve?

Amnesia between sessions
Without it, an AI starts every conversation from zero, asking again for things you already told it.
Generic answers
Once it remembers your preferences, it can give advice that fits you — not one-size-fits-all boilerplate.

How does it actually recognize things?

Classify the type
This sentence has a name, a date, a number, a preference… first sort the info into categories.
Extract the value
In "budget is five thousand", the entity is "budget" and the value is "five thousand".
Decide whether to keep it
Not every word is worth saving. One-off, irrelevant details get filtered out; stable, important ones go to long-term memory.

How it relates to long-term memory

Entity extraction is the writing; long-term memory is the storage. Extracted entities get organized into structured entries in long-term memory. Next conversation, the AI recalls them, making the experience feel like it knows you instead of starting from scratch each time.

Bottom line: entity extraction is the AI's notepad — it grabs the key facts from a conversation and saves them, so you don't have to repeat yourself.