The new optimisation engine evaluates AI agents against business goals, identifies performance gaps and makes targeted improvements before and after deployment.
Enterprise artificial intelligence company Kore.ai has launched Autoloop, an optimisation engine designed to continuously evaluate and improve the performance of AI agents deployed across business operations.
Announced on October 7, 2026, the tool is generally available as part of the Kore.ai Agent Platform, Artemis edition. It enables enterprises to define operational goals and automatically refine AI agents against those requirements throughout their lifecycle.
Autoloop is designed to address challenges associated with maintaining AI agents after deployment, including inconsistent task execution, compliance failures, rising operational costs and limited visibility into automated decisions.
According to Kore.ai's 2026 Agent Productivity Index, 79% of surveyed enterprises reported reversing an action taken by an AI agent, while 70% experienced failures their teams could not trace.
The company said Autoloop replaces repetitive manual adjustments with an automated process that builds, evaluates, diagnoses and refines agents based on predefined business objectives.
These objectives include task completion, response accuracy, adherence to business rules, customer experience, safety and cost efficiency.
The system assesses proposed changes against multiple performance requirements to prevent improvements in one area from negatively affecting another. For example, reducing operating costs should not compromise an agent's ability to complete tasks or follow established safeguards.
Autoloop incorporates StateTrace, a monitoring capability that records an agent's execution history, including tool calls, task handoffs and changes in operational state.
This information is used to identify where an agent failed to meet its objectives and determine which part of its workflow requires modification.
The platform also uses Kore.ai's Agent Blueprint Language (ABL), which converts agent workflows, business rules and operational restrictions into structured instructions.
By connecting execution records to individual workflow components, Autoloop can make targeted adjustments rather than modifying an entire agent configuration.
The optimisation process continues after deployment, using real-world interactions to identify additional performance gaps and verify subsequent improvements.
Kore.ai Founder and CEO Raj Koneru said enterprises seeking to scale AI must establish systems capable of building, governing and optimising agents while maintaining control over business outcomes.
The company also disclosed that AI agents contribute approximately 6,500 code commits each month to its production software environment, which contains around 2.6 million lines of code. These activities operate under 68 continuously enforced safeguards, according to Kore.ai.
Autoloop builds on the Artemis edition of the Kore.ai Agent Platform, introduced in May 2026 to support the development, governance and management of enterprise AI systems.
The launch comes as organisations increasingly deploy agentic AI across customer service, internal operations and enterprise automation.
Kore.ai has not disclosed separate pricing for Autoloop or independently verified customer performance improvements resulting from its deployment.
Disclaimer: This article may include information derived from interviews, press releases, public statements, research, company communications and other publicly available or third-party sources. Such material may be summarised, paraphrased or contextualised for journalistic and editorial purposes. All rights in third-party content remain with their respective owners.