IBM Links Agent Safety to Business Outcomes

IBM is connecting AI agent operations to customer outcomes, infrastructure context and financial accountability. Tyler Lynch, outbound CTO of IBM Automation Portfolio, outlined an approach that extends agent oversight to financial risk, including emitting telemetry for token usage and revoking API keys when policies are violated. IBM supports Nvidia's Open Agent Safety Platform and is expanding HashiCorp's Infragraph into a knowledge graph shared across FinOps and observability tools. Lynch emphasized measuring AI value through business speed and process improvement, not just token consumption, and advised enterprises to start from desired customer outcomes rather than individual technical roles.