Standing up a production agent means taking it through six stages:
At API & AI Summit, we announced the evolution of Kong Konnect into the AI Connectivity Platform and announced a slate of launches built around that idea. Here's what we launched, where each piece fits, and what you can use today.
An agent that only runs when a person prompts it isn't autonomous. Real autonomy starts when something happens in the world, like a payment failing or a ticket opening, and the agent responds. Webhook Engine joins Event Gateway in Event Management to let teams invoke agents reactively from external events, with the same governance as the rest of their traffic.
An agent needs to discover the MCP servers, tools, models, and other agents it is sanctioned to use, and nothing it isn't. The Agent & MCP Registry, part of AI Management, will give agents that governed list. Konnect Catalog, generally available today, is the platform-wide system of record for people: the place teams go to see every model, registry, API, and event stream across the organization.
Agents need access to sensitive systems, and handing them long-lived credentials is how access sprawl starts. Token Vault is designed so agents can reach the systems they need without holding credentials of their own. It builds on Kong Identity as the shared machine identity layer.
Wrapping an existing API as an MCP server is easy. Producing an MCP server an agent can actually use well takes more work. Context Mesh, now generally available, turns existing APIs into well-designed MCP servers, built on code mode, so agents consume governed tools and context instead of raw endpoints.
AI spend grows faster than anyone's ability to explain it. AI cost management, built on Konnect Metering and Billing, attributes agent and model spend so teams can defend it to finance, and lays the groundwork to eventually monetize it. It is available now to early access design partners.
When an agent goes wrong on turn seven of a conversation, request-level logs won't tell you why. Advanced AI Observability traces the whole conversation and lets teams debug it live. It is available now to early access design partners.
Everything above has to run where your agents actually run. For most enterprises, that is several places at once, and the mix will keep changing.
So we don't ask you to pick one. Run agents in AWS, GCP, or Azure. Run them on-prem or at the edge. Provision models in your data platform, or self-host open-weight models. We think a heterogeneous estate is the right enterprise architecture, and we built Konnect to make it governable.
We call this strategic portability. Many AI gateways and data platforms govern well inside their own environment and make everything outside it harder. Konnect is designed to apply one governance model across all of it, which also gives teams a practical answer on sovereignty, data residency, and lock-in.
Agents reach production when the path from trigger to invoice is governed end to end. Konnect is built to give teams that path on one platform, across every environment they run.
Catalog and Context Mesh are available in Konnect today. To join the Webhook Engine private beta or the early access programs for AI cost management and Advanced AI Observability, [talk to our team](https://konghq.com/contact-sales)talk to our team.
We're thrilled to announce that Kong, the AI connectivity company, has been recognized as a Leader in the Gartner Magic Quadrant for API Management for the seventh year in a row. Plus, Kong was positioned furthest on the Completeness of Vision axis
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
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
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:
An AI agent isn't magic. It's a reasoning engine that makes decisions based on what it knows at a given moment. That "what it knows" is its context — the information available to it right now. Context is everything. Give an agent the wrong context,
