Operations | Monitoring | ITSM | DevOps | Cloud

For every $1 you pay OpenAI or Anthropic, it's costing them about $1.60. AI is running at a loss — so is the whole business model broken? The full bill for AI hasn't landed yet. In this ShipTalk short, Field CTOs Adam and Martin break down the economics of generative AI: why the frontier labs lose money on every prompt, why they'd need to raise prices ~60% just to break even, and the bet the entire industry is making — that inference costs drop fast enough to catch up. Plus the Gartner forecast every CFO should see: by 2028, the AI bill could be bigger than the employment bill.

Open source is the backbone of most software. But should developers trust OSS? npm, PyPI, and Maven don't vet packages before publication, so "publicly available" doesn't automatically mean "safe to pull right now." A secure OSS posture is to trust the ecosystem but verify at ingestion: route packages and dependencies through a controlled layer that runs scanning, age checks, and malware detection before anything reaches a build.

OpenTelemetry instrumentation used to mean hours of manual work, wiring up metrics service by service. Now you can just ask for it. Tell an AI agent what you're trying to observe, something like "add OpenTelemetry so I can track this application's performance," and it turns that into an intent specification, then enriches your existing code with instrumentation to match. OpenTelemetry is open source and on GitHub. Pairing it with an AI agent that actually understands your codebase is what turns "add some metrics" into instrumentation that's useful.

Why Observability Systems Fail AI Code.

We spend thousands of dollars "token maxing" and running endless debugging cycles just to walk our LLMs through a problem. But is the AI actually failing, or are we just withholding the right environment? Giving your AI assistant its own sandbox to test hypotheses might just be the missing link in your development workflow.

We intentionally introduced a bug into our app to see what would happen. The agent used the CircleCI MCP Server to detect the failed pipeline, diagnose the issue from the build logs, implement a fix, and keep iterating until CI was green.