Operations | Monitoring | ITSM | DevOps | Cloud

#aicoding #softwareengineering #claudecode

AI coding tools are accelerating development velocity, creating a release challenge most teams aren’t equipped for. Without controlled rollout, higher change velocity makes it harder to know which specific release drove the results you’re seeing in production. And when teams use AI, to build AI – LLM apps and AI agents– complexity multiplies. Traditional observability can’t ensure AI agent quality, performance, and cost-efficiency at production scale.

Every team is doing something with AI right now. What that something is, is an entirely different question. And whether that something is successful? Most teams are still figuring it out as they go.

Replace "AI shipped on hope" with an operating model that holds up once real users depend on it. AI quality is multi-dimensional, covering accuracy, tone, safety, and faithfulness to user data, and can't be debugged from outputs alone. Without visibility into what their AI actually did in production, teams miss regressions, reverse-engineer chains by hand, and watch a single bad answer erode trust built over hundreds of right ones.

Modernizing a legacy system serving 20 million devices without users noticing is like replacing a jet engine mid-flight. In this session, YoungJin Jung and Donggen Hong from LG U+ share their 18-month journey transforming a Telco-scale API Gateway from a rigid, proprietary solution into a high-performance, open-source architecture on AWS, and the operational challenges they solved along the way.

In this tutorial you'll learn how to use Uptrace and Grafana together. Uptrace exposes a Prometheus-compatible HTTP endpoint, so you can add it as a data source in Grafana and reuse your existing dashboards without changing metric names or rewriting queries.