What is AI memory?
Understand the difference between a chat transcript, saved facts, and the context an assistant uses to answer your next question.
Understand AI memory 2 min read
AI memory vs. a context window
A context window is the information a model can work with in a request. Memory is stored information that may be retrieved for a later request. A larger window does not automatically give a new chat access to your old work.
Storage and attention solve different problems
Imagine a folder of project notes and the few pages on your desk. The folder is a useful analogy for stored memory. The pages on the desk are closer to the current context. Something can be safely stored and still be absent from the information used for an answer.
When a chat loses a detail, first check whether the detail was supplied or retrieved. Buying access to a larger model context will not fix a missing connection or an uncaptured conversation.
Build the next request around the next task
A design critique needs the current design, the intended reader, and the constraints. It rarely needs every abandoned brainstorm. Start with the latest decisions and add the original discussion when the assistant needs to inspect the reasoning.
Keep long reference material outside the brief when possible. Name it and explain when to consult it. This makes omissions easier to notice and reduces the chance of an old draft being mistaken for the current one.
TAKE THIS WITH YOU
A compact opening request
An illustrative template. Replace the details with material you are allowed to share.
The next task is to review the signup copy. Use the current draft below and these confirmed constraints. Consult the linked planning conversation only for unresolved questions. Before editing, list any information you still need. Do not assume an earlier draft is current.
USING THIS WITH DRAY
Keep the context. Check the source.
A connected agent can consult stored Dray context through its tools. Whether it calls those tools and uses the result correctly remains something to check.
Dray is in beta. Available sources and connections depend on your setup. Read the data notes before granting access.
Sources & next reads
Product details are grounded in the guides below. Example projects and prompts are illustrative, not customer results.