Quickstart

Your first request in five minutes.

Airelay speaks the OpenAI API. If your code already calls OpenAI, you change the base URL and the key — that's it.

Make your first request

Create an account and a gateway key

Call a free model — no provider key needed

The curated free models work immediately, up to 20 requests a day per organisation.

terminal

curl https://ai.noviqent.co.uk/v1/chat/completions \
  -H "Authorization: Bearer $AIRELAY_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemini-2.0-flash",
    "messages": [{ "role": "user", "content": "Say hello from Airelay" }]
  }'

Connect your own provider key

Add an OpenAI, Anthropic, Google, Azure, Bedrock or other provider key under Provider keys, set up Direct Debit under Billing, and every model that provider hosts is available as provider/model.

Use the SDK you already have

Any OpenAI-compatible client works — only the base URL and key change.

Python — openai

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://ai.noviqent.co.uk/v1",
    api_key=os.environ["AIRELAY_KEY"],
)

reply = client.chat.completions.create(
    model="openai/gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

Node.js — openai

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://ai.noviqent.co.uk/v1",
  apiKey: process.env.AIRELAY_KEY,
});

const reply = await client.chat.completions.create({
  model: "anthropic/claude-3-5-sonnet",
  messages: [{ role: "user", content: "Hello" }],
});

Add a routing policy

Stay up when a provider is down: reference a policy instead of a model.

app.py

# In the dashboard: Routing policies → New → type "fallback"
#   name: primary-with-backup
#   1. anthropic/claude-3-5-sonnet
#   2. openai/gpt-4o

reply = client.chat.completions.create(
    model="policy/primary-with-backup",
    messages=[{"role": "user", "content": "Hello"}],
)

report a bad answer

# Every response has an X-Airelay-Request-Id header.
curl https://ai.noviqent.co.uk/v1/feedback \
  -H "Authorization: Bearer $AIRELAY_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "request_id": "<X-Airelay-Request-Id>", "reason": "empty answer" }'

Switch on guardrails

In the dashboard, open AI → Plugins and enable PII redaction, a prompt guard, the semantic cache or a token budget for your whole organisation. Nothing in your code changes — the next request is already protected.

See every AI plugin →

Publish one of your own APIs

Add a service, a route and a plugin, and your API is live behind the gateway.

API gateway

# Dashboard → API Gateway → Services → New
#   name: orders     url: https://orders.internal-api.example.com
# → Routes → New:   path /gw/<your-org>/orders
# → Plugins → key-auth, rate-limiting (60/minute)
# → Consumers → New "partner-a" → Credentials → API key

curl https://ai.noviqent.co.uk/gw/<your-org>/orders \
  -H "apikey: <partner-a's key>"

Connect an MCP server

Give your agents a tool through a gateway URL you control.

Agents & MCP

# Dashboard → Agents → Directory → pick a server → Connect
# Your connection gets its own gateway URL:
https://ai.noviqent.co.uk/mcp/<connection-id>

# Point any MCP client that speaks Streamable HTTP at it, with
#   Authorization: Bearer <your Airelay gateway key>
# or let the client discover OAuth 2.0 sign-in (Noviqent SSO).

Ready when you are

Create an account, grab a gateway key and send your first request.

Get started free Read the quickstart