What’s New
- Fine-Grained Metric Detection: Support sub-minute metric detection using Kafka for more real-time and accurate anomaly detection
- Auto Escalate Incident: Support auto escalation based on instance level anomaly score
- Outlier Detection: Support string outlier detection in log for data quality issues and more
- Model Drift Detection Algorithm: Support minute level sampling and apply aging algo for CDF distribution aggregation for more realtime detection
- IFTrace SDK: Developed IFTrace SDK for collecting LLM trace
- Deploy AI Observability Product:
- Support model and monitor set up
- Support model drift workbench
- Support workbench RCA timeline
- Support LLM trace viewer and workbench
Enhancements and Improvements
- Metadata Optimization: Optimizing global view and causal metadata return size to increase UI loading
- UX Improvement: Updating problem investigation page UI
- Infrastructure Component Upgrading: Upgrade Cassandra to Cassandra 4
