- Perceive brokers, serving as a single supply of fact to assist mitigate the dangers related to shadow AI.
- Validate every agent earlier than launch by testing for safety, resilience, and coverage compliance to make sure they meet your requirements earlier than going stay.
- Keep management with real-time guardrails that preserve brokers working inside accredited boundaries.
Safety testing, validation, and risk modeling ought to be integrated into growth pipelines, Kyndryl said. “Moreover, runtime protections reminiscent of anomaly detection, guardian brokers, and fast isolation capabilities might help comprise incidents earlier than they escalate. By making safety and governance foundational fairly than treating them as afterthoughts, organizations can confidently scale agentic AI, understanding that dangers are proactively managed, and belief is maintained with clients, companions, and regulators,” Kyndryl said.
The brand new service is simply one of many platforms the seller presents to handle AI brokers. Final 12 months Kyndral launched its Agentic AI Framework. That bundle presents an orchestration system constructed to deploy and handle autonomous, self-learning brokers throughout enterprise workflows in on-prem, cloud, or hybrid IT environments, in keeping with the corporate.
Specialised brokers are deployed to assemble IT info, reminiscent of information evaluation, compliance checks, incident response or service desk ticket decision. Over time, brokers study from information and outcomes to enhance decision-making and adapt workflows autonomously, and an orchestration engine parses that information to let enterprise methods alter to altering circumstances in actual time, Kyndryl said. The platform defines what actions brokers can and can’t do, mainly setting coverage throughout the enterprise.
