Kyndryl unveils enhanced AI framework for secure enterprise scale

Kyndryl has announced significant enhancements to its Agentic AI Framework aimed at supporting enterprises in scaling artificial intelligence projects with a focus on security, compliance, and human governance.

The enhancements introduce a new core to the framework capable of orchestrating, securing, and scaling AI agents across teams and systems, while providing a base for further agentic functionalities. These include tools for analysing and ingesting code, policies and data; an agent builder designed to help teams deploy mission-critical AI agents; and an enterprise-specific workforce model for effective human-digital agent collaboration.

Framework details

Among the new features is the Kyndryl agentic ingestion capability, which analyses organisational code, relevant policies, and business data to build and deploy AI agents focused on concrete business outcomes. Alongside this, the agent builder tool enables the design, testing and deployment of agents tailored to meet security and regulatory requirements associated with mission-critical processes.

The framework features an enterprise-specific model to guide how employees interact with, and are supported by, digital agents to advance productivity and innovation within set business parameters. Security has been highlighted as a foundational consideration, with a 'secure by design' methodology intended to ensure transparency, compliance and adherence to governance protocols for autonomous agents and operations.

Industry application

Kyndryl states that the Agentic AI Framework is currently helping customers accelerate returns on investment within verticals such as government, banking, insurance and manufacturing by reducing the lag between design and deployment of AI solutions.

"With decades of mission-critical infrastructure expertise, unique intellectual property and our AI consult methodology, Kyndryl is poised to lead our customers through this paradigm shift toward agentic AI," said Kyndryl Chairman and Chief Executive Officer Martin Schroeter. "Our differentiated approach blends agents within complex environments and empowers organisations to scale AI throughout their operations as they move to become AI-native."

The company reports that a quarter of its recent engagements now involve AI-related elements, including data architecture and cloud services. Kyndryl's approach is supported by its history of operating mission-critical systems and providing AI-native workflows at enterprise scale.

Partnerships and collaboration

To further develop its framework, Kyndryl is expanding its collaborations through global alliances. This includes a programme with Google Cloud focused on building 100 AI agents in 100 days, as well as partnerships with universities worldwide to foster research and education for the next generation of AI professionals.

Kyndryl is also working with ecosystem partners to integrate its Agentic AI Framework into joint solutions for customers. In the insurance sector, for example, the company is implementing agentic AI-driven actuarial solutions to automate regulatory filings and compliance processes, as well as to provide real-time decision support. In government, Kyndryl is developing and deploying AI agents that operate across multiple departments, such as tax and licensing, immigration, and benefit distribution. These agents incorporate institutional knowledge, work alongside civil servants, and proactively address the needs of both citizens and government employees.

For banking customers, Kyndryl is collaborating to automate and streamline manual client onboarding processes through embedded AI agents that manage application processing, validation, and external vetting, with the aim of reducing onboarding times and improving client engagement.

Technical capabilities

The core of the Agentic AI Framework includes features for dynamic orchestration, security, and scalability of AI agents across a customer's systems. Input from the agentic ingestion tool incorporates analysis of source code, operating policies, data, and insights. The framework is designed with built-in transparency and compliance measures, including so-called 'guardian concepts' aimed at keeping operations both autonomous and auditable.

Kyndryl's future workforce model defines the roles AI agents are expected to play within an organisation and outlines how they should interact with human employees. Company specialists apply the model to identify new professional requirements and skill sets necessary for effective human-agent partnership.

The agent builder tool provided by Kyndryl uses reference architectures and an agent catalogue to allow enterprises to deploy AI agents for activities such as code development, automated testing, and the automation of complex business operations. The tool ensures agents are brought online with regard to compliance and security standards, and that they are fit for mission-critical environments.

Kyndryl continues to engage with customers and ecosystem partners to extend the adoption and reach of its Agentic AI Framework in varied sectors.