Trace detail must explain AI spend without making telemetry unaffordable.
SREs need request-level evidence, while telemetry storage must remain selective and tenant-safe.
Who you’d be doing this for
“Finance asked why yesterday’s token bill jumped, and I’m still matching trace IDs in a spreadsheet.”
Pekka Hansen · Site Reliability Engineer
Runs Kubernetes services that call AI models and is accountable for reliability and cloud spend.
What is at stake
SREs can see rising token spend but still rebuild request evidence outside their telemetry workflow. You must weigh investigation speed against tenant-safe, affordable data retention.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- request-level answers vs bounded cardinality
- investigation speed vs tenant isolation
- pilot learning vs durable schema commitments
- cost control vs diagnostic detail
Why Grafana
Grafana Cloud already receives the traces that SREs use to investigate AI-backed services.
Written with these in mind
Not your kind of problem? 24 more at Grafana, or browse every organization.
This is the setup. The work is inside.
Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.