HeronSentry
HeronSentry is a standalone observability plane for AI Agents: capturing runtime events via standard OTLP to deliver call chain traces, cost aggregation, alerts, and performance exports without depending on other suite products.
HeronSentry
Observability PlatformStandalone AI Agent observability plane: tracing, costs, alerts, and performance analysisWhat Agents do, how well they perform, and what they cost—visible at a glance via standard OTLP.
- OTLP
- Open protocol; semantic fields mapped via profiles
- Full Ingestion
- Unsampled by default
- Synchronous Ingest
- Persisted within request lifecycle
- Standalone
- Deployable with zero suite dependencies
Typical pains
- Troubleshooting broken Agent workflows relies on raw logs without distributed call chains
- Costs remain invisible until month-end invoices arrive after budgets are blown
- No visibility into whether specific Agents are healthy or experiencing abuse
Core value
Pinpoint performance bottlenecks for each agent
When an agent slows down or fails, identifying whether the cause is latency, errors, or output hitting the length limit is difficult. HeronSentry rolls up response latency, TTFT, error rate, truncation rate, recent models and providers in AULO's agent detail view.
Identify which agents drive token and host spend
When costs surge across agent clusters, locating the responsible workloads takes significant effort. HeronSentry groups token usage and spend by model, host machine, and agent in AULO's resource overview, making resource-intensive hotspots immediately visible.
Break down actual AI costs for each project
When delivering business initiatives, attributing LLM expenses across different agents is difficult. HeronSentry attributes project tokens and spend to each model and agent, splitting tokens into input, cache-hit, and output. Budget utilization data is provided by PathPilot.
Get notified promptly on runtime anomalies
When error rates or costs surge, on-call teams risk missing critical incidents. Error rates, cost spikes, ingestion backlogs, and frequent tool calls trigger alerts. Alerts appear on the AULO alerts page, and AULO notifies the responsible owner.
Set alert thresholds to fit your environment
Static rules cause alert fatigue or missed incidents across stages. Create alert rules with custom thresholds and severities in AULO, switch built-in rules on or off, and filter by source, while HeronSentry evaluates them.
Capabilities
Ingestion
Tracing & topology
Cost & performance
Alerting
Does / does not
- Standard OTLP ingestion: handles traces, metrics, and logs at differing processing depths
- Distributed call-chain tracing: reconstruct Agent reasoning, tool calls, and model interaction topology
- Granular token and cache cost accounting: split cache read/write for real-time cost attribution
- Multi-dimensional alerting: error rate, cost spikes, backlog, and anomalous calls
- Agent reliability metrics: TTFT, truncation rate, and tool-rejection analysis
- Analytics and audit data export: metrics and audit feeds for OwlAudit and operators
- Does not serve as a generic enterprise full-text log search cluster (focuses on Agent-runtime OTLP traces and structured logs)
- Does not replace host/network infrastructure monitoring (focuses on Agent application and model-interaction semantics)
- Does not execute active runtime interventions (read-only observability and alerts; blocking belongs to OwlAudit / NodalOS)
By role
Standard OTLP ingestion avoids framework lock-in; coexists with existing observability stacks without claiming backend reuse
OwlAudit pulls operational telemetry exports from this service
Clear cost attribution and measurable ROI for enterprise Agent deployments