Debugging LLM apps with an AI-powered assistant - Opik Virtual Learning Series
👏 Learn how to debug your LLMs with AI assistance!
This is a hands-on workshop series from Comet led by Claire Longo. In this session, we will guide you through an interactive session in Opik, discuss best practices for debugging LLM systems, and facilitate live Q&A.
As developers, we spend countless hours crawling through logs, comparing traces, and guessing which prompt, rule, or model call caused the issue.
But what if AI could help us debug AI?
That’s exactly what we built with Opik Assist — an AI-powered assistant that helps you analyze, debug, and optimize your LLM application traces faster and smarter.
With Opik Assist, you can:
🧠 Ask natural language questions about your current trace and spans
⚡ Automatically identify performance issues and bottlenecks
🛠️ Get debugging recommendations for failed or slow operations
📈 Analyze behavioral patterns across traces
💰 Understand the cost implications of every LLM call
In this talk, we will show how to implement a debugging workflow using Opik Assist in Comet Opik. We will demonstrate how to log traces, uncover insights with Opik Assist, and tune your prompts based on the insights. If you're building AI Agents and passionate about quality, this is the session for you!!