Add Cencori to an Existing Product

Use this guide when you already have a product and want Cencori in the request path. You do not need to rebuild auth, rewrite your database, or start over with create-cencori-app.

The job is simple: create a Cencori project, pick a model that works in that project, put a Cencori key on your server, and point your AI calls through Cencori.

Do these steps in order:

The fastest first test is a catalog or free model that is already enabled for your project. After that works, switch to your preferred production model or BYOK provider.

Before touching code, choose one model ID in the dashboard and use only that model for the first request.

Good first-test choices:

gpt-4o
claude-sonnet-4.5
gemini-2.5-flash

If the dashboard offers a free, managed, or catalog model for your project, use it first. That removes provider-key setup from the first test. If you want to use your own OpenAI, Anthropic, Google, xAI, or custom provider account, add that provider in Project > Providers before testing.

Create a secret key in Project > API Keys and add it to the server environment where your AI call runs:

CENCORI_API_KEY=csk_...

Keep csk_... keys out of browser code, mobile bundles, and NEXT_PUBLIC_* variables. For a Next.js app, put Cencori calls in a route handler or server action.

Most existing products use one of these paths.

Change the API key and base URL. Most OpenAI-compatible clients append /chat/completions automatically, so the base URL should stop at /v1.

import OpenAI from 'openai';
 
const openai = new OpenAI({
  apiKey: process.env.CENCORI_API_KEY,
  baseURL: 'https://api.cencori.com/v1',
});
 
const response = await openai.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello from my existing app.' }],
  user: 'user_123',
});
 
console.log(response.choices[0]?.message?.content);

Python is the same idea:

import os
from openai import OpenAI
 
client = OpenAI(
    api_key=os.environ["CENCORI_API_KEY"],
    base_url="https://api.cencori.com/v1",
)
 
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello from Python."}],
    user="user_123",
)
 
print(response.choices[0].message.content)

Install the SDK:

npm install cencori

Then call Cencori from server-side code:

import { Cencori } from 'cencori';
 
const cencori = new Cencori(); // reads CENCORI_API_KEY
 
const response = await cencori.ai.chat({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello from my existing app.' }],
  userId: 'user_123',
});
 
console.log(response.content);

Use a route handler when your frontend needs to call AI without exposing the secret key:

// app/api/chat/route.ts
import { Cencori } from 'cencori';
import { NextResponse } from 'next/server';
 
const cencori = new Cencori();
 
export async function POST(request: Request) {
  const { message, userId } = await request.json();
 
  const response = await cencori.ai.chat({
    model: 'gpt-4o',
    messages: [{ role: 'user', content: message }],
    userId,
  });
 
  return NextResponse.json({ content: response.content });
}

Your frontend calls /api/chat; your server route calls Cencori.

If code is confusing the issue, test the dashboard setup with cURL:

curl https://api.cencori.com/v1/chat/completions \
  -H "Authorization: Bearer $CENCORI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [{"role": "user", "content": "Say hello in one sentence."}]
  }'

If this works, your Cencori project, key, and model routing are good. Any remaining problem is in your application wiring.

Open the project dashboard and check Logs after the test request. A successful integration should show:

• Request timestamp
• Model and provider
• Input and output tokens
• Latency
• Cost
• End-user ID if you passed user or userId

If no log appears, your product is still calling the provider directly or the request never reached Cencori.

For production, repeat the same setup in your hosting provider:

1. Add CENCORI_API_KEY to production environment variables.
2. Confirm the production project has provider access for the selected model.
3. Redeploy the service.
4. Send one production smoke-test request.
5. Confirm the request appears in the production project logs.

Use separate Cencori projects or keys for development and production if you want clean logs, limits, and spend tracking.

• Quick Start — new app or first Cencori test
• Chat API — native and OpenAI-compatible endpoint reference
• Authentication — supported headers and key formats
• Providers — supported model/provider IDs