Operations | Monitoring | ITSM | DevOps | Cloud

The race on AI is heating up; the next generation of vibe coders and prompt engineers are entering the job market as we speak, and AI is the hottest line item on most IT budgets this year. Building the next great home automation software or adaptive learning platform with cutting-edge machine learning is great and all, but like all great software, it needs to start with the plumbing.

A few days ago I hosted a live conversation with Pranav, co-founder of Chatwoot, about issues his team was running into with LLM observability. The short version: building, debugging, and improving AI agents in production gets messy fast. There's multiple competing standards for default libraries for LLM observability. And many such libraries like OpenInference which claim to be based on OpenTelemetry don't strictly adhere to it's conventions.

AI assistants like Cursor and GitHub Copilot are fun to play with. They autocomplete code, refactor functions, and occasionally argue with you about whether you really needed that semicolon. But the moment you ask them to do something grounded in your repo, say, “start work on JIRA-123”… you hit a wall. They don’t know your branching conventions. They don’t know how your team links issues.

With the influx of AI tooling, it’s never been easier to contribute to open source communities. These tools are capable of gathering context quickly, “understanding” repositories faster than ever before. They provide instant summaries about repositories that, previously, would have meant reading lines and lines of code. They can fix bugs in programming languages you don’t know, and ultimately allow more contributors to get involved, which (almost) every open source project wants.