If you've used KAi before, you'll notice the difference immediately. The enhanced version has deeper access to your Konnect environment and new tools for debugging, analytics, and configuration management:

• - **Active tracing and debugging:** KAi can now create debug sessions that capture real traffic on your routes. Ask "Why is my /api/checkout endpoint slow?" and KAi will set up a tracing session, filter for high-latency requests, and summarize where time is being spent. No more manually configuring debug sessions or parsing trace output.
• - **Analytics queries**: "Which consumers are generating the most 5xx errors today?" "What's the request volume on my payments service this week?" KAi now queries your analytics data directly — filtered by time range, status codes, consumers, services, or routes — and surfaces patterns you'd otherwise dig through dashboards to find.
• - **Deeper configuration access**: KAi can now traverse your entire gateway configuration: control planes, services, routes, consumers, consumer groups, plugins, and vaults. Ask about dependencies ("What plugins affect this route?"), audit configurations ("Show me all rate limiting plugins across production"), or investigate specific entities.
• - **Contextual documentation search**: When you need configuration guidance or troubleshooting steps, KAi searches Kong's documentation and returns answers tailored to your question — not just links to pages.

The new KAi is more capable because it's backed by a structured set of tools that can query and interact with your Konnect environment. Instead of pattern-matching on your question and hoping for the best, KAi now has explicit access to:

• - Gateway entity APIs for configuration queries
• - Debug session creation and active tracing
• - Analytics APIs for traffic and error analysis
• - Documentation search for contextual guidance

Even better, security and access control are taken care of. All queries respect your permissions. When KAi works for *you*, KAi only sees what *you're* allowed to access.

The capabilities powering enhanced KAi aren't locked inside the Konnect console. We've built them on the [Kong Konnect MCP Server](https://developer.konghq.com/konnect-platform/konnect-mcp/)Kong Konnect MCP Server—and that server is available for you to use directly.

The Model Context Protocol is an open standard for AI assistants to interact with external systems. Our MCP server exposes the same tools KAi uses, so you can connect from:

• - **IDEs**: VS Code, [Cursor](https://cursor.directory/mcp/kong-konnect-3)Cursor, JetBrains (IntelliJ, PyCharm, WebStorm)
• - **AI Assistants**: Claude, GitHub Copilot, any MCP-compatible client
• - **Terminal**: Claude Code CLI for command-line workflows

Ask your IDE copilot about your Kong configuration. Debug a slow endpoint without opening the console. Query analytics from your terminal while triaging an incident. The MCP server makes your API platform accessible from wherever you're already working.

Here's where it gets interesting. The MCP server isn't just for human-in-the-loop workflows — it's infrastructure for your own AI agents.

MCP is an open protocol. Any agent framework that supports it can now interact with your Kong environment programmatically. That means you can build:

• - Operations agents that monitor traffic patterns, detect anomalies, and automatically create debug sessions when latency spikes—then summarize findings in Slack before you've finished your coffee.
• - Configuration agents that audit your gateway setup against security policies, flag misconfigurations, and generate remediation plans across all control planes.
• - Incident response agents that correlate error spikes with recent configuration changes, trace failing requests through your service graph, and surface the five things most likely to be causing the problem.

You could even build Agents that will talk and work directly with KAi for multi-agent management of your AI and API platform. If you're already experimenting with agent frameworks like LangChain, CrewAI, or custom orchestration — Kong is now a tool in your agent's toolkit.

While we plan to expand this significantly over the next several months, KAi and the MCP server share this toolset today:

• - **Gateway entities**: Query control planes, services, routes, consumers, consumer groups, plugins, vaults
• - **Debugging**: Create debug sessions with active tracing, filter by latency/path, analyze bottlenecks
• - **Analytics**: Query API requests by time range, status codes, consumers, services, routes
• - **Documentation**: Search Kong docs for configuration guidance and troubleshooting

KAi is already available in your Kong Konnect console. Look for the assistant icon and start asking questions — you'll notice the new capabilities immediately.

If you just want to start working with the Konnect MCP server:

1. -

   Generate a Personal Access Token from your [Konnect settings](https://cloud.konghq.com/global/account/tokens)Konnect settings

2. -

   Add the MCP server to your client using your regional endpoint:

• -

  US: https://us.mcp.konghq.com/

• -

  EU: https://eu.mcp.konghq.com/

• -

  AU: https://au.mcp.konghq.com/

3. -

   Start querying your API infrastructure from your IDE

Full setup instructions for all supported clients are in the[ ](https://docs.konghq.com/konnect/ai-assistant/) [documentation](https://developer.konghq.com/konnect-platform/kai/)documentation.

## What's Next

This is a beta release, and we're actively developing new tools and capabilities. Because KAi and the MCP server share the same infrastructure, improvements benefit both—new tools we add to the server automatically become available in KAi.

We want your feedback. What workflows would save you the most time? What's missing? Let us know as you put this into practice.

*The enhanced KAi and Kong Konnect MCP Server are available now in public beta. *[*Explore the documentation to get started*](https://developer.konghq.com/konnect-platform/kai/)*Explore the documentation to get started**.*