Ask plain-language questions about logs and errors
Ask 'why did checkout errors spike at 2pm?' and an assistant writes the query, runs it and explains what it found.
The use case
Writing log queries in ES|QL or similar languages slows down anyone not fluent in them. Elastic's AI Assistant for Observability takes a plain question, builds and runs queries against logs, metrics and traces, and explains the results with the query shown so it can be checked. It can also draw on your runbooks and knowledge base. This helps support engineers and developers investigate without waiting on an SRE. The person still reads the query to confirm it covers the right time range and service before acting on the answer.
Example workflow
- Enable the assistant and connect a model
- Ask a question about a service or time window
- Review the generated query and results
- Refine the question if the scope is wrong
- Share the confirmed finding with the team
Tools
- Elastic Observability
- Elastic AI Assistant
Risks and guardrails
- Queries can silently use the wrong time range or index
- Logs sent to a hosted model may include personal data
- Answers summarize; check raw events for key decisions
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Elastic docs: AI Assistant for Observability: www.elastic.co/docs/solutions/observability/ai/observability-ai-assistant