It’s somewhat philosophical talk about what’s next. What’s next in the agent world and, you know, I just published a book about the LadyBugDB-based LPG memory and how to model the memory in a proper graph and all those things.

But right now is the question, actually what to do with it? And why do the agents need memory? Yeah, for sure, agents need to remember.

But what do they need to remember and what do they want to do with it? It’s a good question, and yeah, we have a lot of ontologies and data models of memory that somehow mimic human memory, with an episodic events graph, semantic knowledge, complex semantic relations, semantic graphs, and hypergraphs. You can find all this in my book.

But one big missing piece is the cognitive process, right? So, how to actually extract all these ontology structures from the outside world, and it’s still the question that we need to solve.

And I don’t think classical LLMs will give us a straight answer. So it’s not a question of one-shot prompts or some hybrid LLM pipelines, we need something more powerful and more classical, please.

But my thinking these days actually goes to something beyond of the memory. And no, it’s not the context graphs, because actually the memory is more advanced than the context graph and actually you could read it also in my LangGraph book about the proper graphs and the memory and how they actually replace all of this in something much better than the classical context graphs.

But today we will talk about the world models. Actually, why do agents need memory? Why do humans need memory? And you know that we have an internal model of the world around us in our heads, and we accept signals from outside the world that are processed and calculated according to that model.

So it’s how humans function. And we have the concept in our head that describe external world and it’s a big philosophical topic of idealism, if the concept exists separately and what we observe, is it the real thing or only the concept in our head?

But we will not go to this topic because it’s not Friday and my grappa hasn’t been delivered yet. But the simple question of why the agents need memory and what they want to remember?

And yeah, for sure, we already talked that we have the conversational memory to have better conversations with humans. We have the proper graphs and social memory to organize better conversations and interactions between the agents.

But agents need something more than memory. Agents need to form, using their memory, their own world view because they need to understand what’s actually around, what they can do about it, and what you ask them.

Get full access to Sovereign Agentic AI (Volodymyrs View) at volodymyrpavlyshyn.substack.com/subscribe