Systems | Development | Analytics | API | Testing

Traditional observability was built around requests. A request comes in. A service processes it. A response goes out. You trace what happened in between. That model works well for APIs. It breaks down for AI. With agents, a single user request can trigger multiple model calls. An agent might invoke a tool, call another agent, query an MCP server, retry a model, hit a guardrail, and call another tool before finally producing a response. And all of this happens non-determinstically.

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.

Every data platform decision has a default setting, and the default is wait. Not because leaders think the current stack is great. Because “we’ll modernize next year” feels responsible. It sounds like discipline. It reads like you’re protecting the budget. Here’s the part that never makes it into that conversation: waiting is not the free option. It’s a spending decision, and it renews every month whether or not anyone signs off on it.

Across teams, organizations, and industries, people are starting with AI instead of the problem they want to solve. As a result, AI outputs from different tools: This is a process failure, and it's accelerating. Human-in-the-loop (HITL) AI is a framework that integrates human oversight directly into the machine learning lifecycle. Rather than relying on fully autonomous systems, HITL uses humans and machines collaboratively to train models, evaluate outputs, and handle complex decision-making.

If you only read the headlines this year, you’d think AI makes shipping good software easier than ever. Yet, this was the year of very public AI-coding incidents. PocketOS’ production database deletion and a Vercel AI agent shipping unverified code are just two examples of AI-authored code shipped with confidence that turned out to be wrong. And, of course, these AI-coding incidents are distinct from Agentic AI orchestration incidents like the HuggingFace hack by OpenAI.

Corporate sustainability has no shortage of ambition. Over the past several years, companies have set increasingly ambitious climate targets, expanded disclosures, and invested significantly in understanding and measuring their environmental impact. But working in the technology infrastructure industry has made one thing clear to me: Targets don't reduce emissions. Decisions do.

Most enterprise AI infrastructure conversations land on the same tension. Builders want self-service: instant access to GPUs, a stable endpoint for a model, a notebook, or a fine-tuning run that starts without a ticket. Platform teams want per-team quotas, identity-aware access, audit trails, tenant isolation, and cost attribution they can defend. Both sides are right, and most platforms make you choose between them because self-service and control get layered onto a base that was designed for neither.

Two questions stall almost every agentic AI rollout: Is it secure? And what is this going to cost us? Both get harder to answer the longer you wait to ask them, and easier to answer than most teams assume. If you've already invested in a robust data analytics platform like Qlik, you're closer to a secure, governed, and cost-effective agentic AI deployment than the market's “rip and replace” narrative suggests.

Generating code has never been cheaper. Validating it is where teams are losing the most time, and every handoff between tools makes it worse.

Green build pipelines lie. Your automated unit suites can execute without a single failure, API contracts can validate cleanly, and browser regression sweeps can pass with pristine checkmarks. Yet the moment a release hits production, software must interact with state. A single unindexed foreign key, an overlooked lock escalation on a high-throughput table, or an uncalibrated default constraint will bring an entire platform to a halt while your CI dashboards remain reassuringly green.