The CARC Developer Platform.

A Runtime SDK, quickstart, and integration patterns for instrumenting clinical AI agents with runtime control — without rebuilding the agent itself.

Open Developer Portal ↗ Request Demo

This page is a product overview. Full documentation lives in the Developer Portal.

What's included

Runtime SDK

Instrument your agent's recommendations, escalations, deferrals, and overrides so they flow into CARC's Runtime Engine as they happen.

Adapters & Event Ingestion

Connect existing agents through the Adapter Registry or stream decision events in, without changing your agent's own code path.

Evidence & Replay Access

Query Decision Context Records and replay agent runs programmatically through the same API the Control Console uses.

Where the SDK sits

Your Clinical
AI Agent

→

CARC
Runtime SDK

→

Adapter
Registry

→

Runtime
Engine

→

Evidence
Engine

→

Control
Console

Quickstart

A representative example of instrumenting an agent call with the Runtime SDK. Exact method names and syntax are documented in the Developer Portal.

Illustrative example

// initialize the runtime client and start a session
const carc = new CarcRuntimeClient({ apiKey: env.CARC_API_KEY });
const session = carc.runtime.startSession({ agentId: 'triage-agent' });

// record a recommendation with its clinical context
const decision = await session.record({
  recommendation: agentOutput,
  context: clinicalContext,
});

// capture clinician intervention as evidence
await session.onIntervention(evidence => carc.evidence.store(evidence));

Developer Portal preview

developers.giggleaiinnovation.com — Illustrative Preview

GuideQuickstart — instrumenting your first agentDocs

GuideAdapter Registry — connecting existing agentsDocs

ReferenceRuntime SDK method referenceSDK

ReferenceEvidence & Decision Replay queriesAPI

Representative Developer Portal structure — full docs live in the hosted portal

Go deeper

API Reference

Runtime API resources and OpenAPI specification

→

Runtime Metrics

Evidence dashboard and Decision Replay

→

Control Console

Runtime sessions, agent runs, and human review

→

OpenAPI Specification

Machine-readable spec for integration tooling

↗

GitHub

Source, examples, and issue tracking

↗

Ready to instrument your agent?

Request a demo to get design partner access to the Developer Platform.

Back to CARC Request Demo