Quickstart

Five steps take you from nothing to live agent telemetry with per-call estimated cost. The example uses Google ADK; every other framework follows the same shape.

1. Sign up

Create an account at app.radicas.io. Signing up creates your organization — the tenant that owns everything you send. Telemetry, cost data, and dashboard access are all scoped to it.

2. Get your ingest key

Onboarding issues your first ingest API key. It is a low-privilege credential: it can only write telemetry for your organization, nothing else. You can reveal, rotate, and revoke keys later from the dashboard — see API keys.

3. Install and initialize the SDK

pip install "radicas[adk]"

import radicas

radicas.init(
    api_key="___INGEST_KEY___",  # sign in to see your real key
    service="pricing-agent",
    feature="flight-pricing",
)

Sign in to see your real key pre-filled in this snippet. The endpoint defaults to https://ingest.radicas.io — no endpoint configuration is needed. service names the deployable unit; feature tags every span for per-feature cost allocation (see features and allocation). Prefer environment variables? RADICAS_API_KEY works too — see the SDK docs.

4. Run your agent

Run your agent exactly as you normally would. radicas.init() wires the OpenTelemetry providers and auto-activates instrumentation for the frameworks it detects; spans, metrics, and logs export to Radicas as the agent runs.

5. See your data

Open the dashboard and go to the fleet view — your service appears with its features, token usage, and estimated spend. See the fleet manual for what each panel means, or the live ingest status checker if nothing shows up.

Next steps

• Using LangChain, CrewAI, or another framework? Pick your guide in integrations.
• Already run an OpenTelemetry Collector? Point an exporter at Radicas instead — see bring your own Collector.
• Understand what happens to the data: start with the three layers.