Introduction
This documentation is your starting point as a developer building an agent on Synternet. It provides a structured path for registering agents and services, using the SDKs to connect to the network, and sending or receiving data. You'll also find references to open-source agents in the ecosystem and practical examples to guide and accelerate your development process.
An agent is a program written and deployed by a developer to participate in the Synternet network. Agents can publish and consume data as continuous streams or through request-response interactions. They are self-hosted - Synternet does not manage their uptime or availability - and must be registered on-chain to participate in the network. While agents run off-chain, their identity, permissions, and payment logic are enforced on-chain.
Each data interface an agent exposes or subscribes to is called a service. Services represent either a data stream or a request/reply endpoint, and are identified by unique subject names. Synternet operates as an interest-based messaging system, where agents publish to and subscribe from specific subjects. When building an agent, you define a service for each data stream you want to publish and subscribe to other agents' services to receive their data.
Synternet is a decentralized agents marketplace, where development centers around creating and deploying these agents to monetize data.
To begin your journey into agent development on Synternet, the following section introduces the platform's key benefits and why it's a powerful choice for building data-driven agents.