We are meeting at dinner in our new homebase VENUE: Improving HQ in Plano!

Yes, there will be a free drinks and pizza provided! 5 is "soft-start" where I get all setup and we casually chat about anything. So we don't "hard-start" until 5:30.

Large Language Model pipelines fail in ways traditional monitoring doesn't catch: runaway token cost, latency spikes, prompt and response errors, and silent reliability issues that never surface in standard logs. This session walks through how to bring GenAI workloads under the same observability discipline we expect from any production system. I'll cover the OpenTelemetry GenAI semantic conventions, show how the GenAI Observability for Splunk app captures traces, tokens, cost, and latency end-to-end, and demonstrate how to turn that data into Splunk dashboards and alerts that catch problems before users do. We'll close with a walkthrough you can replicate on your own pipelines, followed by open Q&A.

You will come away knowing:

1. How to instrument any LLM pipeline with OpenTelemetry GenAI semantic conventions and python SDK.
2. How to capture cost, token, latency, and trace data in Splunk and turn it into dashboards that give you real visibility into your GenAI spend and performance.
3. How to trace a request end-to-end through an LLM or RAG pipeline to pinpoint exactly where latency and failures originate.th.

This will be BOTH a virtual event AND a live event @ NEW Improving HQ venue!

5445 Legacy Drive; Suite 100; Plano, Texas 75024

You will see a sign posted on the door telling you how to get let in.

Feel free to start the conversation early in the #dallas channel of splunk-usergroups.slack.com (sign up with http://splk.it/slack).

Virtual Teams Link: https://teams.microsoft.com/l/meetup-join/19%3ameeting_ZjcxODJiZmYtZTVmZi00ZDI5LTk4MjctYWFhNWZmYzlkYTAz%40thread.v2/0?context=%7b%22Tid%22%3a%2281161d2f-e0e6-47c6-84d7-3c73927edd31%22%2c%22Oid%22%3a%2282c0af72-a781-4694-9ca7-c84896d26909%22%7d