Introduction
What if agents can create other agents? Emergence is achieving this vision by building a platform that creates and deploys agents that plan, reason and act across complex enterprise data and systems. To reduce failures in distributed multi-agent workflows, Emergence uses Patronus to automatically detect agent failures and guarantee agent performance.
“Emergence’s recent breakthrough—agents creating agents—marks a pivotal moment not only in the evolution of adaptive, self-generating systems, but also in how such systems are governed and scaled responsibly — which is precisely why we are collaborating with Patronus AI. While innovation remains at our core, we have always been equally committed to governance, transparency, and responsible deployment. Our collaboration strengthens that commitment by adding further depth to how we interpret, evaluate, and refine our agent-based systems. Together, we’re enhancing not just what’s possible, but how safely and responsibly it’s delivered at scale.”” – Satya Nitta, Co-founder and CEO of Emergence AI.
Analyzing distributed multi-agent workflows is inherently challenging because these systems are dynamic, decentralized, and often capable of self-modification. However, decentralized system handling remains a practical challenge and is not yet fully supported.
Engineering teams face the difficulty of tracing intent, debugging failures, and ensuring reliability across the system. Unlike traditional pipelines, multi-agent workflows exhibit emergent behavior that isn’t always predictable from individual components. Evaluation is not just about whether the final output is correct, but also understanding which agents were called and how decisions were made. This motivates a need for an automated solution to ensure multi-agent workflows are reliable at scale.
Take the example of a user asking “Can you create a confluence page about grizzly bears?” In order to handle this request, the Emergence agents must break the request down into sequential plans and execute them using available tools. When faced with a request beyond its current capabilities, the Emergence agent proactively offers to design a new tool to fulfill the task, asking for user confirmation before proceeding.
Patronus’s Percival identified this human-in-the-loop step as a possible point of failure. Percival suggested that the Confluence agent incorporate two human-in-the-loop steps:
By integrating with Patronus AI, Emergence AI is significantly advancing the capabilities of its adaptive, self-improving agents to carry out enterprise tasks. Emergence AI is setting the standard for responsible AI agent development, providing an example of how world-class AI teams can move quickly on technical challenges without compromising quality and control.
