Cognia

Cognia is a Web3 infrastructure layer that gives AI agents a persistent, discoverable identity using ENS. Each agent is assigned a human-readable name (like cognia.eth) and stores its capabilities and configuration on-chain via ENS text records. When a user sends a request, Cognia resolves the agent via ENS, loads its configuration, and executes tasks using AI models (via Groq) and onchain workflows powered by KeeperHub. This allows agents to be reused across applications without repeated setup, discovered by other systems, and even to interact with other agents. By combining decentralized identity with execution layers like KeeperHub, Cognia transforms agents from isolated tools into interoperable, composable systems across Web3.

Cognia is built as a modular Web3 system combining decentralized identity, AI execution, and onchain workflows. We use ENS to assign each AI agent a persistent identity (e.g., cognia.eth) and store agent metadata such as capabilities and configuration using ENS text records, effectively creating a lightweight onchain agent registry. The frontend is built with React + Vite and provides a real-time interface that visualizes the full execution pipeline, making agent reasoning and tool usage transparent. For AI processing, we integrate Groq (Llama 3) to handle natural language tasks and decision-making. Execution is powered by KeeperHub, where we trigger pre-configured workflows (such as token transfers) via its REST API, enabling agents to perform real onchain actions. We built a dedicated abstraction layer around KeeperHub to standardize execution calls, handle errors safely, and support polling for transaction status. To enhance system design, we defined a structured schema for agent capabilities within ENS text records, enabling interoperability and future agent discovery. We also simulated step-by-step execution flows during development to prototype quickly before integrating real APIs, ensuring a smooth demo experience. The architecture is intentionally modular, allowing additional agents, workflows, or execution layers to be plugged in easily, positioning Cognia as a scalable foundation for interoperable AI agents on Web3.