Hermes agent has support for multiple AI memory providers. I have covered a lot of services which are officially supported and also recently started exploring the ability of traditional SQL AND NoSQL databases to be a good memory system for Hermes agent.
You don't have to have a database with the vector search support. Instead a traditional SQL engine is more than enough for Hermes agent.
Why? Because by default there are 2 markdown files like user.md and memory.md files that hold the memory for the Hermes agent.
Take an example of SQLite database which I am currently using for the hermes agent. By default the database of the hermes agent is called holographic, which is SQLite database file. But if you want to create your own that is also supported.
SQLite makes data storage simple. And even better is how you store the memories from the Hermes Agent. Some of the time such storage of the files in the database file is simple.
And then making apps around your memories from this file is even more easier. You can even create the diagrams, charts etc out of such facts, preferences too. I think the use cases are endless.
I covered this in my video pointed above. You can check that out and setup the SQLite as a memory for Hermes agent.
