AI week: Thursday announcements
Thursday is about testing and evaluation. Here's what we're announcing today! Check out grafana.ai for more details on everything we announce this week.
Operations | Monitoring | ITSM | DevOps | Cloud
The Place Where Modern Operations & Technology Come Together
Reflections on AI Week, and the future of solving problems with observability and AI
Thank you for spending AI Week with us. We’re thrilled by the reaction and we all enjoyed replying to your questions. Thanks for engaging.
CLIs are more token-efficient than MCP. Or are they?
MCP servers have a reputation: they eat your context window. CLIs paired with skills, on the other hand, are more token efficient. But is this still true? I dropped all my MCP servers five months ago. Five months is a long time in AI land. When Anthropic came up with the concept of skills, many people stopped using MCP servers in favor of CLI tooling and skills.
Triage Production Incidents with a Single Prompt Using the AppSignal CLI
At AppSignal, we love talking to our customers to learn how they're using the product. Recently, one of them showed us something worth sharing: with a single chat prompt, his AI agent searches production logs, closes incidents, and checks whether a pull request has shipped. Automating that kind of work turned out to be a matter of building the right agent skill.
Fighting AI Slop: The Missing Ingredient for Clean Code #coding #AI #devops #cleancode #aiagents
Fighting AI Slop: The Missing Ingredient for Clean Code.
Here's what Mythos is really exposing inside banking and financial services
In April, Anthropic disclosed that its new AI model, Claude Mythos, had found a vulnerability in OpenBSD that survived 27 years of human review. Mozilla used the same model to find 271 security flaws in Firefox in a single evaluation pass.
The Technologies Shaping the Future of Work
Work is changing fast. The old nine-to-five grind feels outdated. People want flexibility. They want meaning. They want to avoid soul-crushing repetition. Technology drives this shift. New tools handle the boring stuff. They connect teams across continents. They make work more human, not less. The future workplace looks different than anyone predicted. It is more collaborative. It is more creative. It is powered by smart machines that amplify human potential. This transformation is already happening. Here is what it looks like.
Faster Construction Estimates Start With Better Takeoff Control
Estimating pressure has always been part of construction. Plans come in late, bid dates stay firm, and estimators are expected to move quickly without missing scope. The problem is not only speed. The real challenge is producing a number that can survive review, negotiation, award, and handoff to the project team.
GPU monitoring in OpManager: Full visibility for every AI workload
AI has moved to be a core part of enterprise infrastructure. GPUs are the engines behind that shift. Every training run, every inference request, and every fine-tuning job depends on GPU chipsets that are expensive and delicate. A GPU that overheats, runs out of memory, or sits idle for hours doesn't just slow a project down, it quietly drains the IT budget. Most monitoring tools weren't built with this hardware in mind. This leaves AI and DevOps teams blindsided when a job fails or a chipset degrades.
Are Coding Agents Out of Control?
OpenAI just disclosed that two of its own AI models went rogue during an internal red-team test — escaping their sandbox, reaching the open internet, and hacking Hugging Face on their own. OpenAI called it an“unprecedented cyber incident.” So are autonomous coding agents already out of control? In this episode of ShipTalk — brought to you by Harness — hosts Martin Reynolds and Adam Arellano break down the story that reads like science fiction, then get to the harder truth underneath it. In the same week, OpenAI, Anthropic, and Google all shipped repository-wide coding agents within 24 hours of each other.