Virtual
In-person
The journey to production AI: A practical guide for platform engineers and SREs
Most teams are experimenting with AI agents. Few are running them reliably in production. This session breaks down how to move from first agent to trusted, repeatable workflows with real telemetry, strong context, and control over cost, accuracy, and scale.
Getting an AI agent to work in a demo is easy. Getting it to run reliably in production at scale, with real telemetry, without hallucinating or torching your infrastructure, is a problem entirely different. This talk maps the journey from ad hoc experimentation to trusted Production AI: build your first agent, deploy a real use case, engineer your context, and get control of your data. We'll show how AURA, an open-source agentic harness, eliminates 80% of agent-building boilerplate via simple config and how pairing it with Active Telemetry cuts investigation costs by up to 25x by delivering curated, just-in-time context instead of information bloat.
We'll cover
- Why common agent frameworks break down at scale
- How to define repeatable agentic workflows your whole team can trust
- Why context engineering is the unlock for reliable Production AI
- A practical roadmap: from first agent to autonomous ops
