Introduction to Observability

Monitor inference performance, investigate requests, and enforce protection rules

Overview

Bud AI Foundry Observability provides a single workspace for understanding inference behavior across models, deployments, projects, and users. It combines metrics, request-level diagnostics, and enforcement rules so teams can monitor service health and respond quickly.

Key Features

Inference-Centric Metrics Track requests, latency, token usage, TTFT, and throughput over configurable time windows.

Request-Level Debugging Drill into individual requests with metadata, payload details, and trace-ready identifiers.

Rules & Blocking Controls Manage IP, country, user-agent, and rate-based rules with active/inactive state and lifecycle actions.

Consistent Filtering Apply time range, text search, and scoped filters to narrow investigations quickly.

Core Workflows

1. Monitor trends on the Metrics tab.
2. Investigate anomalies from the Requests tab.
3. Protect traffic using Rules tab controls.
4. Iterate by adjusting filters and repeating analysis.

Getting Started

Quick Start

Explore metrics and requests in minutes

Observability Concepts

Learn key objects, signals, and tabs

First Investigation

Run a practical investigation flow