Dash0 Acquires Lumigo to Expand Agentic Observability Across AWS and Serverless

Last updated: March 3, 2026

Key Concepts

While getting started using Agent0, it helps to understand how it works under the hood.

Overview

This section covers the core architectural components — the specialized agents, the tools they use, and how context flows from your application into the conversation.

Agents

Agent0 uses a multi-agent architecture. When you submit a query, an automatic routing layer (the Generalist) analyzes your request and delegates it to the most appropriate specialized sub-agent. Each agent is optimized for a specific domain:

AgentSpecialty
SeekerDetects anomalies, pinpoints root causes across traces, logs, and metrics for faster incident triage.
OracleExplains, translates, and optimizes PromQL queries with precision, eliminating guesswork.
PathfinderLeads OpenTelemetry instrumentation and service onboarding in minutes, not weeks.
ThreadweaverBuilds cause-and-effect narratives from traces, revealing bottlenecks and dependencies.
ArtistDesigns custom dashboards and alert rules to visualize complex systems clearly.

Tool Invocations

Each agent has access to a set of tools — API calls and data retrieval operations used to build responses.

Every request triggers one or more tool invocations behind the scenes.

These are fully transparent: click any tool call in the response to inspect exactly what the agent did, including the queries it ran and the data it retrieved.

Integrated Context

Agent0 can automatically infer context from wherever you are in the application.

Context is injected in two ways:

Integrated Context

  • Button triggers: Certain views (such as alert details) include a dedicated Agent0 button. Clicking it opens a pre-populated prompt with the relevant service, time range, and alert information already loaded.
  • Command palette (⌘K / Ctrl+K): From anywhere in the application, open the command palette and type your question. Agent0 captures the context of the current view — the service, time range, and filters — and uses it to inform its response.

Actionable Tags

Responses include actionable tags — clickable elements embedded in the text.

Clicking a tag (such as a service name, trace ID, or metric) opens a contextual side panel with detailed information, allowing you to drill deeper without leaving the conversation.