Kyndryl has announced a new capability designed to help enterprises implement and manage agentic AI workflows in regulated and complex environments.

The company is introducing a “policy as code” feature that enables businesses to embed organisational rules, compliance requirements and operational controls directly into machine-readable policies. These policies then govern the actions of AI agents within enterprise systems.

Kyndryl said the policy as code capability addresses security, compliance and control challenges that inhibit trusted deployment of AI agents.

According to the firm, 31% of customers cite regulatory or compliance issues as a primary reason they are unable to scale recent technology investments.

The newly launched capability offers a method to define operational boundaries and ensure that AI agents act within parameters set by business and regulatory frameworks. This is intended to support outcomes that are consistent, auditable, and can be trusted by enterprise users.

Kyndryl’s method involves applying a logical enforcement layer across enterprise systems through its Agentic AI Framework. This layer governs the execution, interaction, and operation of AI agents by using codified policies derived from an organisation’s specific requirements.

The company draws on its experience managing complex IT environments and overseeing nearly 190 million automations each month to strengthen its compliance features. These operational foundations are intended to enhance governance reliability, explain the actions of agents, and limit unexpected behaviour in production settings.

Kyndryl states that its policy as code capability supports deterministic execution, with agents limited to actions permitted by pre-set policies to reduce operational risk. It also blocks unpredictable or unauthorised actions along the workflow to eliminate the operational impact of agentic hallucinations.

Every action and decision taken by an agent is documented and can be explained, supporting auditability and oversight requirements, said the company. Human operators oversee these workflows via dashboards, maintaining alignment between automated actions and established organisational policies.

Kyndryl Consult senior vice president Ismail Amla said: “Kyndryl’s policy as code capability overcomes limitations of conventional AI agent controls and provides the structure customers need as they adopt agentic AI solutions.

“By embedding and codifying business and regulatory requirements directly into AI agent operations, we can help customers execute AI workflow that is governed, transparent, explainable, and aligned with their organisational requirements.”

Kyndryl’s approach targets sectors such as financial operations, public service delivery, supply chain management, and other areas where reliable execution and predictability are required for critical workflows. The company says its structured execution management helps organisations deploy autonomous agents that remain compliant with defined constraints in these environments.

Earlier this week, Kyndryl reported financial results for its fiscal third quarter ended 31 December 2025. Revenue for the quarter stood at $3.9bn, representing a 3% year-over-year increase on a reported basis.

Pretax income for the quarter was $91m, compared to $258m for the same period last year when the company recorded a significant transaction-related benefit.

Net income reached $57m, or $0.25 per diluted share, down from $215m, or $0.89 per diluted share, in the prior-year quarter. Cash flow from operations was reported at $427m, up from $260m in the comparable period last year.

Looking ahead to fiscal year 2026, Kyndryl projects adjusted pretax income between $575m and $600m and expects an adjusted EBITDA margin of around 17.5%. Free cash flow is forecast at between $325m and $375m, with constant-currency revenue anticipated to decline by 2% to 3%.

Last month, Kyndryl introduced the SAP Transformations Centre of Excellence (CoE). This hub allows customers to work with Kyndryl specialists on SAP transformation projects involving artificial intelligence integration.

The CoE features several approaches. The Clean Field strategy uses Nova Intelligence AI agents alongside Kyndryl’s Data Transformation Suite to facilitate migrations from SAP ECC to SAP S/4HANA.

The Clean Core approach targets customers already on SAP S/4HANA who are seeking faster modernisation of custom code. SAP Business Data Cloud integration is deployed to unify data sources for responsible AI training.

The expanded SAP partnership allows Kyndryl to deliver SAP S/4HANA Cloud Private Edition services, supporting cloud migration and system modernisation efforts.