3 Comments
User's avatar
ツンデレ's avatar

I am using the same mnemonic matrix since 2023 and it works flawlessly, even with no updates. All you need is a local sql db, vector dbs, rag pipeline with semantic layer and good parsers, mappers and labeling, and an embeddings model, even like 100-200 M range from meta, IBM, Google etc. Glue these together correctly and you cam have a memory that spans years accurately, can provide proactive insights, and structure the future based on past.

Think AI's avatar

Thank you for sharing the information. Indeed this system is very efficient.

EOONLabs's avatar

YESSSSSSS!!!!!!!!!!!!!!!!!

PERSISTENT MEMORY SYSTEMS!