When you add something to Loom, it gets stored in a way that makes it easy to find later. When you search, Loom looks at your memory three ways and combines them into one ranked list of results.

Meaning

Loom understands what your text is about, so a search finds related ideas even when the words are different.

Connections

Loom tracks how papers, claims, and sessions relate, so a good match can pull in its neighbours.

Time

Loom knows when each fact was true, so you can ask about a point in time.
1

Loom reads your question

Your question is turned into the same kind of representation as your stored memory, so the two can be compared.
2

It gathers candidates

Loom collects likely matches from each of the three views above.
3

It combines them

The candidates are merged into a single ranked list, so a result that’s strong in more than one view rises to the top.
4

It follows connections

Loom looks at what the top results are linked to and brings in related pieces that help answer the question.
5

It favours what's current

Newer, still-valid facts are preferred, so you tend to get today’s answer.
6

It fits your budget

Results come back until they fill the token budget you asked for, each with its source.

​
Remembering over time

Loom keeps track of when each fact was true. That means you can ask what your memory looked like at an earlier date, and when two facts disagree, Loom keeps both rather than overwriting one. This is helpful in research, where findings get revised and it matters which version you’re looking at.

Try a search

How to search, with examples and options.