Kong Catalog is the system of record for every interface running across your organization — regardless of which team built it, which cloud it runs on, or which gateway sits in front of it. Where it previously tracked APIs, Catalog now treats AI Models, MCP Servers, Agents, and event streams as first-class entities. Coming soon, each will have its own documentation and versioning support.

The goal is simple: one place where a platform team can answer "what exists, who owns it, and is it safe to build on?" — for APIs and for the AI infrastructure now sitting next to them.

The sprawl problem APIs faced a decade ago is repeating itself with AI, only faster and with higher stakes. As organizations stand up AI Models, MCP Servers, and Agents across different teams and clouds, these interfaces rarely land in a shared inventory that a platform team can trust. And the problem doesn't stop at AI — it gets harder once you factor in the APIs and event streams that agents also depend on as tools and context. Without a common system of record for all of it, platform teams lose visibility into what exists, who owns it, and whether it's safe to build on.

Gartner projects that the average global enterprise will run more than 150,000 AI agents by 2028, up from fewer than 15 in 2025 and only 13% of organizations believe they currently have the right governance in place to manage them. Gartner's prescribed first step for managing that sprawl is building a centralized agent inventory.

• - **Every AI interface, one catalog.** AI Models and MCP Servers join APIs as first-class entities in Catalog, now generally available. Agents joined in beta, with GA landing soon.
• - More coming to every entity type. Documentation and versioning support – already available for APIs – are extending to AI entities next.

**Semantic search, powered by KAi. **Platform teams and builders can ask Catalog questions like "are there any agents that can summarize legal contracts?" instead of hunting across teams and tools.

Today, most organizations are handling this with spreadsheets, tribal knowledge, or point solutions that only cover their own cloud. Cloud-native AI registries are the closest thing to a purpose-built answer, but they share the same limitation: they're scoped to a single provider's ecosystem, which works fine until your organization, like most, ends up running models, agents, and MCP servers across more than one cloud.

Kong is designed to be cloud and gateway agnostic, pairing Catalog with the Developer Portal and gateway runtimes you're likely already running. This results in one inventory across every cloud, rather than a registry that only sees what happens inside a single one.

If you're running Service Catalog today, it moves into sustaining mode: no new functionality is being added to it, but there's no end-of-life date either. You're not forced onto the new experience on any timeline. Integrations and Scorecards aren't carried forward in their current form, but the jobs they did — third-party discovery and quality assessment — are being rebuilt around APIs and AI interfaces rather than services.

The new Kong Catalog is GA. Kong Catalog is the governance layer for the AI infrastructure your teams are already building — no forced migration, no registries to maintain, one inventory whether the interface in question is an API, a model, an MCP server, an agent, or an event stream. If you're a Kong Konnect customer, [_learn more about Kong Catalog_](https://developer.konghq.com/catalog/)_learn more about Kong Catalog_ and reach out to your account team if you have questions!