Take Control of the Economics of AI with Kong AI Gateway

Kong expanded its AI Gateway with new cost management capabilities that price AI interactions at the individual request level and attribute spend across models, providers, applications, teams, users, and agents, the company announced. Kong said the gateway sits between AI consumers and models, letting organizations trace costs such as a $27,000 Claude Sonnet charge to a specific customer support agent, returns-and-refunds workflow, region, and cost center. The company framed the release as closing an "attribution gap" in which enterprises can see total AI spend but cannot explain what drives it.

Take Control of the Economics of AI with Kong AI Gateway Alex Drag Head of Product Marketing New AI cost management capabilities help organizations price, attribute, plan, control, and optimize AI spend across models, applications, teams, and agents. AI is getting easier to build with. Paying for it is getting harder to manage. As organizations move from AI experiments to production deployments, AI consumption is spreading across applications, teams, developers, and increasingly autonomous agents. Behind all of that activity are models from multiple providers, each with different pricing structures and consumption patterns. The result is a new operational challenge: enterprises can see what they’re spending on AI, but often can’t explain what they’re spending it on, much less understand the savings being realized by AI initiatives Today, we’re expanding Kong AI Gateway’s cost management capabilities to help organizations take control of the economics of AI — from accurately pricing individual AI interactions to understanding what’s driving that spend, planning and controlling consumption, and ultimately optimizing the value they get from every AI dollar. The AI attribution gap Most AI cost management starts with infrastructure. A provider can tell you that you consumed a certain number of input and output tokens on a particular model. An observability platform can show requests, latency, tokens, and traces. A cloud cost platform can show how spend changes over time. All of that is useful. But businesses don’t operate in tokens. They operate in teams, applications, products, programs, customers, cost centers, people, and increasingly agents. That’s the attribution gap. Imagine an enterprise spends $500,000 on AI this month. Knowing that $300,000 went to one provider and $200,000 to another is useful for accounting. But it doesn’t answer the questions needed to actually manage that spend: - Which applications and AI initiatives are responsible for it? - Which teams, users, or agents are driving consumption? - Are those teams operating within their budgets? - Where is spend growing unexpectedly? - Could some workloads use a less expensive model without sacrificing quality? - Are we getting enough business value from what we’re spending? Without attribution, AI cost visibility tells you what happened. It doesn’t give you control over what happens next. From cost visibility to cost control Kong AI Gateway operates at a unique point in the AI architecture: directly between AI consumers and the models they use. That means the gateway can see the interaction as it happens — including the consumer, application, model, provider, token consumption, and cost. We’re building on that foundation to create a more complete operating model for AI economics: Price Understanding AI spend starts with getting the economics right. There isn’t a universal “cost per token.” Pricing varies by provider and model, but can also vary based on input versus output tokens, cached tokens, context-window thresholds, service tiers, and other provider-specific dimensions. Kong calculates AI cost at the individual request level, creating a consistent economic foundation across models and providers. Attribute Accurate pricing tells you what an AI interaction costs. Attribution tells you why that cost exists. Kong connects AI consumption with the business context behind it, allowing organizations to understand spend across multiple dimensions — from an individual interaction, user, agent, or workflow all the way up to an application, team, program, product, or cost center. Instead of simply seeing: Claude Sonnet → $27,000 an organization could understand: Customer Support Agent → Returns & Refunds Workflow → North America → Customer Experience → $27,000 That distinction turns infrastructure telemetry into business information. Instead of knowing “we spent $27,000 on Claude,” the organization can understand “our Returns & Refunds agent spent $27,000 resolving 18,400 customer cases, at an average AI cost of $1.47 per case.” Now AI cost can be connected to business activity and ultimately business outcomes — creating the foundation for budgeting, optimization, and understanding AI ROI. Plan Once AI spend has business context, organizations can begin managing it against the way they actually operate. Budgets can be aligned with teams, applications, programs, or other organizational dimensions rather than simply with cloud accounts or model providers. That gives teams ownership over their consumption while giving platform, FinOps, and finance teams a consistent view across the organization. Control A budget isn’t particularly useful if you only discover you’ve exceeded it when the invoice arrives. Kong helps organizations move from reactive cost reporting toward proactive cost governance. By combining real-time AI consumption with budgets and policies, organizations can identify anomalous consumption, understand where spend is trending, and take action before unexpected usage becomes unexpected cost. And because Kong operates inline with AI traffic, cost controls don’t have to stop at an alert or dashboard. Policies can be applied directly to AI consumption. Optimize The goal isn’t simply to spend less on AI. It’s to get more value from every AI dollar. Kong AI Gateway provides multiple ways to improve the economics of AI workloads, from semantic caching that eliminates unnecessary model calls, to prompt compression that reduces token consumption, to intelligent routing that directs workloads to the most appropriate model based on cost, performance, and quality. But optimization only matters if teams can see the impact. Kong measures the savings generated by each optimization, showing where savings come from and how they accumulate over time. It also identifies additional opportunities to reduce costs and estimates the savings that could be captured, helping teams continuously improve AI economics without limiting adoption. Optimization therefore becomes an infrastructure capability rather than something every development team has to solve independently. One economic layer across your AI infrastructure There’s another reason we believe the gateway is the right place to manage AI economics: models will change. The model that’s right for a workload today may not be the right model six months from now. Organizations will adopt new providers, negotiate new commercial terms, deploy private models, and continuously rebalance workloads based on price, quality, latency, and business requirements. Your business structure doesn’t change every time your model does. The Customer Experience team is still the Customer Experience team. Your claims-processing application is still your claims-processing application. Your coding agent is still your coding agent. By separating the business attribution model from the underlying AI infrastructure , Kong gives organizations a consistent economic control layer across providers. Change the model. Change the provider. Change where it runs. The business context remains intact. AI governance has to include economics As AI becomes a fundamental part of enterprise infrastructure, governance can no longer mean security alone. Organizations need to govern who and what can access AI. They need visibility into what AI systems are doing. They need policies around how models, APIs, and tools are consumed. And they need to govern the economics. That’s where we’re taking Kong AI Gateway: toward a unified control point where enterprises can manage the security, reliability, and economics of AI connectivity. Because ultimately, the question isn’t how many tokens your organization consumes. It’s what those tokens accomplish — and whether they’re worth what you’re paying for them. Introduction to OWASP Top 10 for LLM Applications 2025 The OWASP Top 10 for LLM Applications 2025 represents a significant evolution in AI security guidance, reflecting the rapid maturation of enterprise AI deployments over the past year. The key up Michael Field How to Master AI/LLM Traffic Management with Intelligent Gateways As businesses increasingly harness the power of artificial intelligence AI and large language models LLMs , a new challenge emerges: managing the deluge of AI requests flooding systems. This exponential growth in AI traffic creates what could be Kong How the Rise of Agentic AI is Transforming API Development and Management The world of artificial intelligence is undergoing a seismic shift, with the emergence of agentic AI redefining the landscape of API development and management. As businesses and developers navigate the complexities of digital transformation, unde Kong AI Token Cost Management: Why AI Spend Gets Out of Control and How To Fix It This post is based on Kong's webinar on token cost management. Watch it below, or keep reading for the breakdown. Token cost management is how a business tracks, controls, and governs what it spends on AI model usage, the same way it already track Kong Kong A2A and MCP Metrics: Visibility and Governance for AI Tool Adoption at Scale When an organization deploys AI agents at scale, high uptime and low latency are an important baseline. However, Platform owners and business stakeholders could be flying blind on several fronts: The Insights Gap: Non-technical stakeholders have li Amit Shah Your Multi-Agent System Is Only as Reliable as Its Context Layer Multi-agent workflows live and die on context. Every agent-to-agent call and every agent-to-tool call is either a retrieval — fetching information the agent needs — or a mutation — changing state that downstream agents will depend on. At prototype s Hugo Guerrero The Architecture Decision Your Multi-Agent System Will Live With Multi-agent systems are, at their core, context distribution systems. Every agent in your workflow is a consumer and producer of context. The interesting architectural questions are all about how that context moves. Two operations drive everything: Hugo Guerrero Securing Enterprise AI: OWASP Top 10 LLM Vulnerabilities Guide Introduction to OWASP Top 10 for LLM Applications 2025 The OWASP Top 10 for LLM Applications 2025 represents a significant evolution in AI security guidance, reflecting the rapid maturation of enterprise AI deployments over the past year. The key up Michael Field How to Master AI/LLM Traffic Management with Intelligent Gateways As businesses increasingly harness the power of artificial intelligence AI and large language models LLMs , a new challenge emerges: managing the deluge of AI requests flooding systems. This exponential growth in AI traffic creates what could be Kong How the Rise of Agentic AI is Transforming API Development and Management The world of artificial intelligence is undergoing a seismic shift, with the emergence of agentic AI redefining the landscape of API development and management. As businesses and developers navigate the complexities of digital transformation, unde Kong AI Token Cost Management: Why AI Spend Gets Out of Control and How To Fix It This post is based on Kong's webinar on token cost management. Watch it below, or keep reading for the breakdown. Token cost management is how a business tracks, controls, and governs what it spends on AI model usage, the same way it already track Kong Kong A2A and MCP Metrics: Visibility and Governance for AI Tool Adoption at Scale When an organization deploys AI agents at scale, high uptime and low latency are an important baseline. However, Platform owners and business stakeholders could be flying blind on several fronts: The Insights Gap: Non-technical stakeholders have li Amit Shah Your Multi-Agent System Is Only as Reliable as Its Context Layer Multi-agent workflows live and die on context. Every agent-to-agent call and every agent-to-tool call is either a retrieval — fetching information the agent needs — or a mutation — changing state that downstream agents will depend on. At prototype s Hugo Guerrero The Architecture Decision Your Multi-Agent System Will Live With Multi-agent systems are, at their core, context distribution systems. Every agent in your workflow is a consumer and producer of context. The interesting architectural questions are all about how that context moves. Two operations drive everything: Hugo Guerrero Securing Enterprise AI: OWASP Top 10 LLM Vulnerabilities Guide Introduction to OWASP Top 10 for LLM Applications 2025 The OWASP Top 10 for LLM Applications 2025 represents a significant evolution in AI security guidance, reflecting the rapid maturation of enterprise AI deployments over the past year. The key up Michael Field How to Master AI/LLM Traffic Management with Intelligent Gateways As businesses increasingly harness the power of artificial intelligence AI and large language models LLMs , a new challenge emerges: managing the deluge of AI requests flooding systems. This exponential growth in AI traffic creates what could be Kong How the Rise of Agentic AI is Transforming API Development and Management The world of artificial intelligence is undergoing a seismic shift, with the emergence of agentic AI redefining the landscape of API development and management. As businesses and developers navigate the complexities of digital transformation, unde Kong AI Token Cost Management: Why AI Spend Gets Out of Control and How To Fix It This post is based on Kong's webinar on token cost management. Watch it below, or keep reading for the breakdown. Token cost management is how a business tracks, controls, and governs what it spends on AI model usage, the same way it already track Kong Kong A2A and MCP Metrics: Visibility and Governance for AI Tool Adoption at Scale When an organization deploys AI agents at scale, high uptime and low latency are an important baseline. However, Platform owners and business stakeholders could be flying blind on several fronts: The Insights Gap: Non-technical stakeholders have li Amit Shah Your Multi-Agent System Is Only as Reliable as Its Context Layer Multi-agent workflows live and die on context. Every agent-to-agent call and every agent-to-tool call is either a retrieval — fetching information the agent needs — or a mutation — changing state that downstream agents will depend on. At prototype s Hugo Guerrero The Architecture Decision Your Multi-Agent System Will Live With Multi-agent systems are, at their core, context distribution systems. Every agent in your workflow is a consumer and producer of context. The interesting architectural questions are all about how that context moves. Two operations drive everything: Hugo Guerrero Ready to see Kong in action? Get a personalized walkthrough of Kong's platform tailored to your architecture, use cases, and scale requirements.