Quickstart

Get started with Eridian in under five minutes. This guide walks through SDK installation, project setup, and your first inference call through the orchestration layer - including semantic caching, routing, and observability fields on the response.

1. Get Your API Key

Sign up for an Eridian account and create a project. Your API key is available under Settings → API Keys in the production dashboard. Keys are prefixed with eridian_sk_ and scoped to a single project.

Minimum scope for this quickstart: inference:write, models:read.

2. Install the SDK

# Python
pip install eridian-sdk

# TypeScript
npm install @eridian/sdk

# Go
go get github.com/eridian-ai/sdk-go

All official SDKs target https://api.geteridian.com/v1. See SDKs for OpenAI-compatible configuration.

3. Make Your First Inference Call

from eridian import Eridian

client = Eridian(api_key="eridian_sk_...")

response = client.inference(
    model="auto",
    messages=[
        {"role": "user", "content": "Summarize the key risks in this contract."}
    ],
    project_id="prj_legal_001",
    features=["pii_redaction", "semantic_cache"],
)

print(response.choices[0].message.content)
print(f"Model: {response.eridian.route}")
print(f"Latency: {response.eridian.latency_ms}ms")
print(f"Cost: ${response.eridian.cost_usd:.4f}")
print(f"Cache hit: {response.eridian.cache_hit}")

TypeScript equivalent

import { Eridian } from "@eridian/sdk";

const client = new Eridian({ apiKey: "eridian_sk_..." });

const response = await client.inference({
  model: "auto",
  messages: [{ role: "user", content: "Summarize the key risks in this contract." }],
  projectId: "prj_legal_001",
  features: ["pii_redaction", "semantic_cache"],
});

4. What Happened

You made an inference call through the Eridian orchestration layer. In production:

• Routing selected an optimal model (eridian.route)
• Semantic caching checked for similar prior prompts (eridian.cache_hit)
• PII redaction masked sensitive entities before provider dispatch (eridian.pii_redacted)
• Observability logged the full span for audit and cost attribution

5. Verify in the Dashboard

Open the observability dashboard in your workspace to see request volume, latency, and cost for your project.

Next Steps

• Configure routing policies
• Set up budgets and rate limits
• Enable PII redaction defaults
• Upload RAG documents

Choose a vertical

Inference is the shared substrate. Product workflows live on typed routes:

Troubleshooting

See Authentication for scopes and Inference for the full request schema.