Introducing agentic traces

Ever wondered what’s happening under the hood when your agent is processing a request? Now you can see it all in action, without writing a single line of monitoring code. That’s right - total observability comes built-in with every agent you deploy on Blaxel!

What are Traces and why should you care? 🤔

Think of traces as a GPS for your agent’s journey. Every time your agent processes a request, it creates a trace that shows you exactly what happened, when it happened, and how it all fits together. It’s like having x-ray vision into your agent’s thought process!

Each trace is made up of spans - think of them as the building blocks of your agent’s journey. A span shows you:

When something starts and ends

What went in and what came out

How different parts of your agent work together

Details about LLM calls, tool usage, and processing steps

Features that make our traces special

Our traces come with some pretty neat features out of the box:

Zero setup required: Deploy from our console or SDK, and traces are automatically enabled

Complete visibility: See everything from LLM calls to tool executions in one clear view

Smart sampling: We automatically save traces for 10% of your executions to help you monitor performance

Debug mode: Need to trace a specific execution? Just add debug=true to your API call

Enterprise-ready from day one

For our enterprise users, we’ve got even more. Our traces use the OpenTelemetry standard, which means you can easily export your data to your favorite monitoring tools like Datadog.

Ready to get started? Head over to your Blaxel console and check out the traces for your agents. We can’t wait to hear how this helps you build better, more reliable AI applications!

Bringing unprecedented clarity and efficiency to AI development

Developer experience is at the heart of everything we build at Blaxel. Our traces feature exemplifies this commitment, providing developers with powerful observability tools that make testing, iterating, and releasing AI applications faster and more reliable than ever before.

Our developer-first approach sets a new standard for AI agent development. By making engineering best practices the default, we’re reshaping how teams build and deploy AI applications. This is just the beginning - we’re committed to creating tools that

Have questions or feedback? We’d love to hear from you! Drop us a line and let us know how you’re using traces to debug and optimize your agents.