This section explains how memejob AI agents are structured at a system level, separating data ingestion from agent behavior to ensure modularity, auditability, and extensibility.
Memejob AI agents are composable by design and are built around two core pillars that operate in tandem:
The Data Layer
Responsible for sourcing, normalizing and exposing on-chain, off-chain and oracle-based data.
The Agentic Layer
Encapsulates agent behavior and decision-making workflows.
This separation ensures that data sourcing and normalization can evolve independently from agent logic, reducing coupling and enabling safer iteration.
Together, these layers enable modular, extensible agents, that can adapt strategies, integrate new data sources and evolve behavior without tight coupling between data ingestion and decision logic.
You can find implementation details of agent execution and SDK-level capabilities in the Agent Framework section.