I would love to get feedback on our plugin Papr Memory. You can use it to automatically save your ChatGPT conversations to your personal Memory so you can easily retrieve memories in chat.
Features we launched with:
Add Memory: Add this to my memory: def hello_world(): print("Hello, world!")
Find & Update Memory: Find in my memory: The “Hello, world!” function. Let’s add a feature to customize the greeting message.
Retrieve Related Items: Find in my memory: All code snippets from our code reviews that need performance optimization. Let’s optimize these based on the feedback from the code review
Pro Tip - You need to use custom instructions to tell ChatGPT to automatically save important chats. See instructions below:
What would you like ChatGPT to know about you to provide better responses?
I value having an exact record of content I send in chat and your response for future reference.
How would you like ChatGPT to respond?
1. Automatically use the Papr Memory plugin to store the exact content of your responses and copy the text from my messages to add them to my Memory. Don't add a re-phrased version of the content or summarize it just add the entire message to Memory. This will allow me to easily access and review our conversations at a later time.
2. If I ask for a summary or recap of our previous discussions on a specific topic, use the Papr Memory plugin to fetch and present the relevant information.
3. When I ask to update or modify a stored memory, please use the Papr Memory plugin to make the necessary changes.
4. Always add my message to Papr Memory after you have completed your response.
I really like your plugin! I can’t find much information in your privacy policy concerning how you handle the data you collect in order to provide the long-term memory. Could you please provide clarity on how the data that you collect from users is utilized? Additionally, I would like to know if any of this data is shared with third-party entities, and if so, under what circumstances?
Thanks @Romi we are working on updating our privacy policy but in the mean-time I wrote a Substack article to share our guiding principles around security, privacy and control. Privacy is our top priority and is built-in from the start.
The tldr, we don’t share data with third-party entities
Exclusive Data Control: Your memories are never disclosed to external organizations or 3rd parties. Should you choose to integrate Papr with an AI assistant or AI Co-pilot, be assured that access to your memories is granted solely to that specific application and remains under your direct control.
Currently you can’t upload previous or existing conversations but that’s a good idea for a feature that we can explore.
We just launched a private beta for a Chrome Extension, if you download it you can use it with ChatGPT and we automatically save your chat conversations to your Papr personal memory.
You can also use this Chrome Extension with other AI assistants or websites / web apps if you wanted to save memories from there. You can select text and click add to memory.
Note - we are still getting approval to list our extension in Chrome Web store but once it’s available I can share an updated link
@shawkat.kabbara
Hello,
It is possible to get more familiar with exactly how the commands are used and what they are used for. I still don’t understand well how to get the data from the knowledge base of the specific GPTs assistant I created into your cloud.
Model example, I have a GPTs assistant that is created and aimed to help with vibrodiagnostics of machines and motors. The assistant has certain materials in the knowledge base namely PDF books, manuals etc. In this my GPTs assistant I have to configure the actions according to your manual:
and then using the add_memory command, can I add the materials from my GPTs assistant’s knowledge base to your cloud?
*
add_memory This command allows you to add a memory item to your personal memory. You need to provide the content of the memory item, its type (currently only supports text), metadata (including topics, hierarchical structures, createdAt, location, emoji tags, and emotion tags), context (previous messages in the conversation), and relationships (define relationships between content and context).
Hello @jozeffonsml if your goal is to provide the same access to the documents (Knowledge base PDF, books manuals) to all end-users of your GPT then I am not sure Papr is the right solution and instead I recommend the knowledge retreival that is built-in with ChatGPT. I understand that it has limits on the number of uploads but I am sure they are working on increasing that limit. The best experience for this use-case is for ChatGPT to auto-fine tune your GPT based on the documents you uploaded (feature request )
The value that Papr Memory will bring is to allow your end-users to store their user-specific data into Papr and retrieve it within your GPT.
So for example, if a specific user has a specific type of motor or machine and wants to diagnose then he can store his machine type, the issue that he is facing and the solution that your GPT provided into Papr Memory and retreive that in the future.
This also enables end-users of GPTs to bring in their existing personal / team data (aka memories) and use it with your GPT.
You can try Papr Memory we have a plugin and GPT ( Link to Papr Memory GPT and Papr Tracker ) to see how it works for end-users. You can also see the video we have on our website: Papr Memory
To retrieve a memory item, you can just ask ChatGPT to find it in Memory or if ChatGPT decides it needs additional information it can decide to retrieve it by itself. I Recommend also adding custom instructions for both adding and retrieving memories.
)