Whenever a user submits a prompt that generates a data model, a memory can be created. A memory will include the data model and the code generated to get the data.
You can then save a memory, add a title and a description, and other users will be able to see and use the memory.
This is very useful to ground your team to use specific tables that are validated and tested.
In your prompts, you can reference memories by using the @ symbol. Using memories in prompts is recommended, and a great way to provide more context to the LLM so it can generate better results.
For example, if a user saved a “Top Customers” memory, and you want to get the top customers in a specific time period, you can use the following prompt:
Copy
Get me @Top Customers for the last 30 days
You can also reference multiple memories in a single prompt. For example, if you want to get the top customers and the top products, you can use the following prompt:
Copy
Get me @Top Customers that purchased @Top Products