Memory Overview

Embra has a dynamic memory that can store and interrelate information.

There are different core types of memory, and you can customize sub-types for each core type to benefit your work.

Meetings: Meetings will show up here once you integrate your calendar or have Embra record meetings outside of your calendar.
Companies: When you meet with companies or email them, they will show up here automatically. You can also manually create them or import data by giving it to Embra and asking it to create the companies.
People: When you meet with people or email them, they will show up. You can also manually create them or important data.
Work: Default sub-types include task, project, goal and milestone. You can create action items from meetings or track work if you set up agents to remember the work.
Documents: Documents include all text-based content. Documents can interconnect all other forms of memories by at-mentioning them (type ”@” in document creation). Documents server as the core free-form way for Embra to remember anything and relate that memory to anything else. As more integrations are made into external applications like Google Docs or Notion, those pages will also show up as Documents.
Extractions: Extractions are structured, smaller pieces of information parsed from other data.

You can type “@” to search for a specific memory to include in the chat context.
- This allows you to work off of that specific memory, search from it, edit it, etc.
If memory is enabled in any chat or agent, you can ask Embra to search its memory for any information or specific records.
You can tell Embra to save data into its memory by providing the data and giving the instruction, either in chat or in agents.

Go to settings by clicking the your account avatar in the sidebar, and then click settings. Once there, a memory page will be near the bottom.
You can add new types of memory, such as types of meetings, types of companies, etc. Adding types enables data filtering and agents automation integrations.