Agent Architecture

AgentNode is a network of interoperable, on-chain agents that work together to streamline and simplify the DeFi experience.

Each agent operates with a defined Objective and a corresponding Action Space, which together shape its decision-making and behavior.

---

🎯 Objective

When deploying a new agent, users define an Objective—a clear, high-level instruction like:

"Allocate 100 USDC daily to the top 5 highest-yielding pools."

These objectives can be updated at any time via a natural language interface, making it easy to adapt strategies on the fly.

---

🛠️ Action Space

Once the objective is set, the Orchestrator Agent scans the entire Action Space—a library of available tools and operations—to determine how best to fulfill the goal.

Using the above example, actions might include:

• Query all active lending pools on Solana

• Supply USDC to Solend

• Withdraw from Solend

• Compare APYs across platforms

• Reallocate based on new yields

This modular design ensures each agent can dynamically compose workflows from available primitives.

---

⏱️ Task Scheduler

For fully autonomous agents, the Orchestrator spins up a Task Scheduler based on the user's objective.

This enables agents to run on a recurring basis—daily, hourly, or in response to on-chain events—without user intervention.

---

With AgentNode, automation isn’t just smart—it’s adaptive, composable, and conversational.