You only ever talk to one thing: the Loom API at https://api.loom.getmetacognition.com. MetaCognition runs and scales everything behind it, so there’s nothing for you to set up. Here’s how the pieces fit together.

The API layer

This is what your requests hit. It reads the files you send, breaks them into searchable pieces, runs your searches, and handles sessions, accounts, and billing.

The storage

This is where your memory lives. It keeps your data safe, makes searches fast, and tracks how your papers, claims, and sessions connect — and when each fact was true. You can read more in Storage and durability.

What a request looks like

Adding memory and searching it are the two calls you’ll make most. Everything else — sessions, accounts, billing — works the same way: one request in, one clear response back.

Ways to connect

REST API

Call Loom from any language.

Python SDK

A small Python client for the same API.