AI Cost Attribution & Optimization for LLM Consumption
AI is rapidly shifting from an experimental feature to a core product capability delivered through APIs. But unlike traditional APIs, LLM consumption has highly variable costs driven by prompt size, output length, model choice, reasoning depth, retries, tool calls, retrieval, and agent workflows. This session explores how organizations can make AI consumption economically sustainable by moving beyond request-based metrics toward AI-specific metering, cost attribution, governance, and optimization. It examines how costs can be attributed across users, applications, tenants, API products, models, and workflows, and how these insights can inform quota design, model access, policy enforcement, and pricing strategy.
Attendees will learn how to approach AI APIs as measurable, governable, and monetizable products, with practical patterns for optimizing LLM usage and building sustainable AI API business models.
Speakers
Anusha Jayasundara is a Senior Technical Lead at WSO2 with over nine years of experience driving innovation in API management, governance, analytics, and cloud-native integration. He has led the design and implementation of key capabilities in WSO2 API Manager, with deep expertise in API Analytics, streaming data integration, and runtime governance. Anusha is passionate about building scalable, secure, and insight-driven API ecosystems that empower organizations to thrive in multi-cloud and hybrid environments.