Neuro San Studio

Neuro San is an open source, data driven library from Cognizant for building multi agent systems. Agent networks are defined in HOCON configuration files with minimal orchestration code, which the vendor says allows fast iteration and modular deployment across tools, models and infrastructure. It is the open source library behind the Neuro AI Multi Agent Accelerator. Requests are routed through a configurable frontman agent that delegates work to specialized agents. Each agent can reason independently, communicate in natural language, call external tools or APIs and collaborate with the others. Tasks can be routed using AAOSA, described as flexible, self organizing agent behavior. Python tools, APIs and third party agents such as Agentforce and Agentspace can be connected. Sensitive inputs are protected with a mechanism called sly_data, and detailed logs let teams monitor agent behavior. The page also encourages vibe coding, testing and scaling of agent networks. The framework is aimed at developers and enterprise teams that want to prototype and scale agent networks without writing the coordination code by hand. The page offers a bring your own key version to try and a link to the GitHub repository. No prices are published for the framework, which sits within Cognizant's AI lab section.