From Zero to Monitoring: Installing Observability in Your AI App in 2 Minutes

For the modern developer, agility is paramount. Building Artificial Intelligence applications often relies on a stack that prioritizes speed and ease of use. It's common to see projects that combine the power of Supabase for databases and authentication, the efficiency of Stripe for payments, and the intelligence of Large Language Models (LLMs) for advanced functionalities. This combination allows independent hackers and small teams to launch innovative products in record time.

However, speed in development shouldn't mean neglecting observability. Monitoring the behavior of your AI application is crucial to ensure its stability, performance, and, of course, to avoid surprises on your bill. The good news is that you don't need to be an infrastructure engineer to have complete visibility. KoraLog was created with exactly that in mind: to offer a simple and quick-to-install observability solution that integrates seamlessly into your development stack.

Installing KoraLog: A Simple Step-by-Step Guide

The KoraLog installation process is extremely straightforward, designed to get you monitoring your application in minutes. There are no complex agents to configure, sidecars to manage, or YAML files to edit. The philosophy is simple: two lines of code, total visibility.

Step 1: Install the SDK

Depending on the language you use, installation is done with a single command:

npm install koralog          # Node.js / TypeScript
# pip install koralog        # Python
# composer require koralog/sdk  # PHP

Step 2: Add one line per LLM call

After installing the SDK, you only need to wrap your existing LLM calls. Your core logic stays untouched, KoraLog sits alongside it, not inside it.

import { Koralog } from "koralog"

const kora = new Koralog({ apiKey: process.env.KORALOG_API_KEY })

// wraps your existing LLM call nothing changes
const res = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: prompt }],
})

await kora.track({ prompt, response: res, model: "gpt-4o" })
// ↑ done. full visibility.

With these two steps, your AI application will already be sending data to KoraLog, allowing you to monitor errors, latency, and costs in real time. The big advantage is that you receive alerts directly on WhatsApp or by email, without needing to open a dashboard or configure complex integrations with tools like Slack although this option is available in more advanced plans.

Why This Approach Works

This vibe-friendly approach from KoraLog means you can focus on what really matters: building and innovating, without worrying about becoming an expert in monitoring infrastructure. Have peace of mind knowing your AI application is under control, with proactive alerts keeping you informed of any anomalies, whether it's a bug, a latency spike, or an unexpected increase in costs.