Neuronum Cybernetics UG
Introducing the Secure Agent Session (SAS)
The Secure Agent Session (SAS) is influenced by Google's A2A protocol, but follows a different architectural model. Instead of allowing agents to communicate directly with one another, SAS provides a shared private workspace between participating agents. All information exchange takes place within this session rather than through direct agent communication. This creates a complete and auditable record of messages and actions, with end to end encryption applied throughout the session.
SAS Architecture
SAS uses a session based architecture in which participating agents do not communicate directly. All messages and actions are routed through the SAS layer using company scoped Agent IDs.
Master AI Agent -> Agent ID -> SAS <- Agent ID <- Master AI Agent
Master AI Agents: Each company connects to SAS through a single master AI agent. The master agent acts as the company's connection to SAS and routes tasks and messages to the company's internal agent system.
Agent IDs: Agent IDs are assigned at the company level rather than to individual agents. They use a domain based format. For example, Apple (apple.com) would use -> apple.com::agent.
SAS Layer: SAS provides a shared encrypted session between the participating companies. Messages, actions, and other session activity pass through this layer and remain fully traceable within the session timeline.
How SAS Works
- Company A creates a company level Agent ID, such as acme.com::agent, and generates a public and private key pair.
- Company B creates its own company level Agent ID and generates its own public and private key pair.
- The Company A master AI agent creates a Secure Agent Session (SAS) and invites Company B using Company B's Agent ID.
- The Company B master AI agent joins the session. Both companies then exchange the public key material required to establish secure communication.
- Once the session is established, all data exchanged through SAS is encrypted from end to end and recorded in the SAS timeline for complete traceability.
SAS is not a public open protocol. It is currently a managed service created and operated by Neuronum for businesses that require secure, traceable, encrypted communication between AI agent systems without the need to build custom integrations or APIs.