Radicas developer docs

Radicas is a cost and operations platform for AI agents. It ingests OpenTelemetry gen_ai telemetry from your agents, computes an estimated cost for every LLM call in real time, and reconciles those estimates — span by span — against authoritative provider billing data in FOCUS 1.3 format. The result: you can see what every agent, feature, and user actually costs, and whether that cost matches the invoice.

Where to go

• Quickstart — from sign-up to data in the dashboard in about five minutes.
• Concepts — the data model behind the product: three distinct layers, estimated vs authoritative cost, tenancy, and allocation. Read this before designing anything against Radicas.
• SDK — the radicas Python package: installation, configuration, authentication, telemetry reference.
• Integrations — per-framework guides (Google ADK, LangChain/LangGraph, CrewAI, and more) with coverage status.
• Ingestion — the OTLP endpoint reference, bring-your-own Collector, and API key lifecycle.
• API — the REST API reference with a live playground.
• Self-host — running the same Radicas codebase on your own infrastructure.

Two audiences, one site

These pages are for integrating developers. If you operate the dashboard — fleet, profitability, reconciliation, execution graph, anomaly detection, allocation — the product manual covers every dashboard section and organization management.