Where the workflow shifted
OpenTelemetry GenAI semantic conventions organize model requests, responses, tokens, tool calls, and errors into portable observability fields.
If every agent tool logs differently, later evaluation, cost review, and client acceptance become difficult to merge.
Tool names are not outcomes
The signal matters when it changes how a team ships, reviews, or recovers work, not when it only names another tool.
Check permissions and failure
- Use provider, model, operation, token, tool, error, and latency as common log fields
- Keep the test narrow: one low-risk task or tool entry before connecting permissions, logs, failure handling, and human takeover to production
What still needs proof
Unstructured logs make model, provider, version, and workflow comparisons unreliable. Keep the original source open so the announcement, the evidence, and this site's interpretation stay separate.