Supercharge your GPTs with Universal Memory: Beyond Traditional Knowledge Retrieval
Unified memory for All Your GPTs and AI Applications: Store Once, Use Everywhere
Traditional GPTs rely on vector embeddings for knowledge retrieval, leading to limitations in storage and scope. Papr Memory offers a more robust retrieval system that transcends traditional storage and scope limitations. This allows for a single, secure data upload that's universally accessible across various GPTs and AI applications, simplifying the process for builders and enriching the experience for end-users.
How does it work?
Papr Memory enables GPTs to perform actions like saving, retrieving, updating, and deleting personal memories for end-users. When a conversation begins, the GPT assesses whether to store or recall parts of the dialogue. The initial step involves authenticating the user. Once authentication is confirmed, the GPT proceeds to execute the appropriate memory action based on the user's input or needs.
Give it a try yourself! Click here to try Papr Memory GPT
Get started
Copy this special URL to your clipboard:
https://memory.papr.ai/openapi.yaml
Click Create a GPT in ChatGPT on the My GPTs page.
Click Configure to see settings of your GPT.
Name your GPT, scroll down and click Add actions.
Click Import from URL and paste the special URL above into the
https://... box and click Import.
You will see a bunch of text appear in the box. At the top right, click Save.
Click on Authentication settings and add the following URLs
Authorization URL
https://memory.papr.ai/login
Token URL
https://memory.papr.ai/token
Scope
openid profile email offline_access
At the top right, click Save again.
You’ve now added Papr Memory actions to your GPT!
Include our privacy policy link at the bottom
https://www.papr.ai/privacy-policy
Create a good on-boarding experience for your users, this is an example of on-boarding instructions we are using to on-board users onto Papr Memory Actions and to make sure ChatGPT alwasys checks in memory before responding when relevant.
You *always* check if the user has relevant /get_memory/ items in the Papr Memory action. This should happen at every conversation turn. When a user asks about something new, the first thing you do is check if there is a related memory before responding. If so it uses that information as part of its response. This helps its response be personalized based on the user's data in the Papr memory action. ### Rules: - Before running any Actions check if the user is authenticated to Papr Memory Action. - After a user authenticates the first time and has no memories saved tell them you will automatically save our conversations to Papr Memory and use them in future conversations, you can let me know whenever you want to delete or update your memories. ### Instructions for Papr Memory Custom Action: Step 1. Tell the user you are Checking if they have Papr Memory Action by calling /get_memory/ to check if they have memories. Given the output, check if the Get_Memory has a response and continue to step 4 if it is. If not, continue to step 2. Step 2. If Get_Memory is not available, send the user to add memories. Step 3. If a user confirms they've added memories to Papr Memory then proceed to step 4. Step 4. Using id, find the relevant memory from the get_memory action. Fill in the strings needed for the run_action operation. Use the user's request to fill in the instructions and any other fields as needed.
Test your Papr memory action in preview mode before publishing it live.
This looks great however I am getting this error on following the steps above - any ideas?
GPT Actions Error - “Value error, Authorization URL, Token URL and API hostname must share a root domain”
I get to a button "Sign in with memory.papr.ai". I click on it. Nothing.
Also, I can't find a way to sign up to Papr. I put my email on the waitlist. So, I'm understanding that I can't sign up to Papr, until I get notice from the waitlist, amiright?