Skip to main content
AI agents are hard to debug. They make dozens of API calls, branch on unpredictable responses, and fail silently in ways that are nearly impossible to reproduce without a full record of what happened. OTAS gives you that record — a structured log of every action every agent took, grouped into sessions, surfaced through a real-time analytics dashboard.

​
What problem OTAS solves

When something goes wrong in an AI-powered system, the first question is always “what did the agent actually do?” Without observability, you’re left guessing. OTAS captures the full HTTP trace for every in-domain and out-of-domain request your agents make, storing path, method, status code, latency, request and response bodies, and custom metadata. You can replay a session, pinpoint the failing call, and fix only what broke — without re-running the whole pipeline.

​
Key capabilities

End-to-end event capture

Log every HTTP request your agents make — in-domain or external — including headers, bodies, latency, and custom properties.

Session-scoped tracing

Group events into agent sessions so you can trace exactly what happened during a single task or run.

Latency percentiles

View daily p50, p95, and p99 latency breakdowns per agent and path, rendered as charts in the dashboard.

Error rate monitoring

Track error counts over the last seven days and drill into individual failing events to understand root causes.

​
Services overview

OTAS is composed of two backend services and a React dashboard:
  • UASAM (port 8000) handles user accounts, authentication, projects, agents, and API key management. It is built on Django REST Framework with PostgreSQL and Redis, and uses PyJWT for both user tokens and agent session tokens.
  • Brain (port 8002) receives event logs from your backend SDK or directly from agents. It stores structured BackendEvent records and exposes analytics endpoints for path timeseries, latency percentiles, session events, and error counts. Celery handles async processing.
  • Frontend (port 5173) is a React 19 + MUI 7 dashboard where you create projects and agents, view session lists, and explore charts for traffic, latency, and errors.
All three components communicate through a set of custom HTTP headers (X-OTAS-USER-TOKEN, X-OTAS-PROJECT-ID, X-OTAS-AGENT-KEY, X-OTAS-AGENT-SESSION-TOKEN, X-OTAS-SDK-KEY) that scope every request to the correct user, project, agent, and session.

​
Next steps

Get started

Start all three services and log your first event in under 15 minutes.

Architecture overview

Understand how UASAM, Brain, and the frontend fit together.

Build docs developers (and LLMs) love

Get started for freeTalk to us