Operations | Monitoring | ITSM | DevOps | Cloud

Every AI agent architecture diagram shows the same five layers: perception, memory, reasoning, action, and feedback. Each layer assumes the one before it worked correctly, and none of them can confirm that once the agent runs against live production data. Runtime context is the sixth layer most designs leave out, and it’s the one that decides whether any of the other five can be trusted.

What 554 developers and engineering leaders told us about AI, agents, and the measurement gap nobody’s closing. Ask a developer if AI made them faster this year, and 84% will say yes. Ask their VP to put a number on it for the board, and 39% will have nothing to show. That gap, not adoption, is the real story in engineering right now.

Education is becoming more connected than ever across classrooms and campuses. Students learn through cloud platforms. Teachers use digital tools to collaborate and engage students. Institutions rely on connected systems to manage research, administration, communications, and campus safety. Technology is not just creating new possibilities for education, but it is also introducing new risks. Behind every device, application, and online account is a piece of valuable information.

Moving the agent off your machine and pointing it at real clusters — read-only, in shadow mode — then climbing a trust ladder toward carefully scoped action. This is the second article in a three-part series on taking an AI SRE agent from a weekend experiment to enterprise production. Part 1 built a local agent on a throwaway cluster: read-only, propose-only, refined against a small eval set, with portable skills and no production write access.

It's been almost exactly one year since we issued our AI mandate here at Honeycomb, and we've been doing some reflection. When we issued our mandate, it's not like we hadn't been using AI. We were the first in the industry to bake a feature powered by AI into our product, way back in May of 2024. Many of us had been experimenting and using these tools in our spare time. But we believe that software is the killer app for AI.

Summary: An API defines how software interacts with a service. MCP defines a standard way for AI applications to discover and invoke tools exposed by a service. They usually work together: an MCP server can sit in front of APIs you already run, turning low-level operations into capabilities an agent can find and use at runtime. Your API may already expose everything an AI agent needs. The harder problem is helping the agent figure out which operations matter for the task it has been given.

Engineering leaders have been claiming to “shift quality left” for years but production remains stubbornly stuck out of reach of software engineers. The realm of production remains mysterious with tools no one has access to and UIs that wouldn’t make sense to engineers anyway. I’ve noticed a small but growing trend of large enterprises hiring Reliability Engineers instead of Site Reliability Engineers. Dropping one word looks cosmetic but I think it points to a much bigger change.

On the product team here at Grafana Labs, we consider AI agents our users, too. That’s why we set out to test how well agents can debug incidents across the full stack, and how much better they perform with Grafana Cloud’s Knowledge Graph vs. using raw telemetry alone. Our early results are promising. In one real incident we replayed 16 times each way, an agent with Knowledge Graph context found the correct root cause 15 times, compared with just once using raw telemetry alone.