Taming the agentic influx: a blueprint for AI business observability (thenewstack.io)

In a recent discussion, Kin Lane, co-founder of Naftiko, highlighted the urgent need for improved "business observability" in AI systems, comparing the current AI surge to the earlier cloud migration era. Lane warns that organizations pouring resources into AI often lack visibility into costs and outcomes, creating hidden liabilities akin to API and application sprawl. The disconnect between engineering and business—where engineers struggle to communicate technical challenges to stakeholders—hinders effective AI management and financial accountability. Lane proposes a solution centered on enhancing technical observability to include business metrics, advocating for a tagging system that embeds structured metadata into API interactions. This approach would allow organizations to answer critical financial questions related to AI spending, such as cost per customer and revenue attribution by product line. By adopting established domain-driven design principles and leveraging FinOps strategies, businesses can gain greater control over their AI expenditures. Lane emphasizes that without this framework, companies risk being overwhelmed by their AI investments, which he predicts could drastically exceed previous cloud costs.