This is the title of Yefim Natis‘s post about agentic computing. “Agents can be the building blocks for composable solutions, and orchestration agents may take on the role of a composer. Autonomy of a software component is not new. The best architectures for complex enterprise systems are compositions of autonomous components, typically interconnected via asynchronous messaging and event-driven communications. Event listeners (sinks) are always autonomous as they are designed to operate in the “mission-impossible” style: without the ability to negotiate with the source. An autonomous component must ensure its internal integrity independent of external events and contexts. Autonomy does not mean that the component does not connect out; it means that it will operate with full integrity regardless of the outcomes of its external connections.

Autonomy’s value does not require the use of AI. Autonomous modularity is the typical choice for complex systems that demand performance at the enterprise/global scale. The organizations that are hesitant to invest in the current AI tools for their mission-critical systems should not delay the architectural transformation to agentic computing. Autonomous modularity and the principles of composable architecture will serve you well in improving your current systems, and will turn out to be a great foundation for the future enterprise-AI computing.” Link