Splunk Observability for AI

[DayaSCanales]

Don’t miss out on an exciting Tech Talk on Splunk Observability for AI!

Discover how Splunk’s agentic AI observability provides complete, end-to-end visibility across your entire AI stack—from large language models to the infrastructure beneath. This session is available now on demand, will showcase how Cisco AI Defense integration helps proactively detect guardrail violations, keeping your AI applications safe, reliable, and effective in production.

Explore Splunk Observability for AI here: 

Get ready to learn how to monitor user and agent conversations with automated evaluations for bias, toxicity, hallucinations, sentiment, and relevance. This insightful session will equip you with the tools to enhance customer trust, improve user experiences, and lower operational costs. Seize this opportunity to build resilient, responsible, and secure AI application stacks!

[DayaSCanales]

Here are a few more resources you might find helpful:

Voice of the Customer:

• Observability for AI 

Documentation:

• Introduction to Splunk Observability for AI 

On - Demand Webinar:

• Agentic, AI-powered Observability: Simplifying Operations & Faster Resolution 

Blogs:  

• A Recap of .conf25 Splunk Observability Announcements 

• Monitor the Health, Performance, and Security of Your AI Application Stack with AI Agent and AI Infr... 

Talk to us if you have any questions:

• Talk to an expert

[LesediK]

You might find the live Q&A interesting. View questions submitted by your peers.

Q. What might make someone a good candidate for Splunk Observability for AI?

A. If you're already running AI applications or have AI infrastructure, whether you're starting out or you're more mature in your AI journey, and if you're facing the same problems around not having any visibility or insights into not only performance, but also the cost, security or quality of responses, you're a prime candidate for observability for AI - AI Infrastructure Monitoring and AI Agent Monitoring.

This is the solution to help you correlate those insights to your other meta data, your logs and traces, have it in one single view, and connect that key business metrics.

Q. If I’m a Cisco customer, can I use Splunk Observability for AI to monitor my AI pods?

A. Absolutely!

If you're familiar with Cisco's AI PODS solution, or if you're not, it is a pre-configured packaged hardware solution for training and inferencing. So, it's an all in one solution using UCS hardware, Nexus switches, a storage provider, and, a number of GPUs coming from Nvidia. The hardware package is kind of the easy button for on-prem inferencing and training.

Splunk Observability, is optimized for that solution. Using AI Infrastructure Monitoring, you can view out-of-the-box dashboard views around all of the technology in the Cisco AI as well as the technology that's coming from what you're building into the AI POD to run your application. Once you deploy the AI POD and configure out your observability, you can be up and running and inferencing in a matter of very little time.

Q. When will Observability for AI be GA?

A. AI Infrastructure Monitoring is generally available today. Head to our docs to learn more and get started. AI Agent Monitoring is in Alpha and will be GA soon. You can sign up for the preview to get started here. 

Q. What's the pricing model if I already have AppDynamics?

A. Please talk to your Splunk Account Rep and to find out what the details are to get started with AI Agent Monitoring. If you're a Splunk AppDynamics customer, you should already have these capabilities, as they became GA in September 2025. 

Q. For the agentic infrastructure monitoring, how is that enabled?

A. The infrastructure monitoring components – all the dashboards and navigator views – is available to all customers. It will appear in your environment when you start sending data in.

Head to the documentation, deploy open telemetry collector to go and collect information from your GPUs or any of the component technologies listed that you're collecting it from. Once you see that data pattern coming in, the dashboards and navigator views will appear in your environment.

Q. Once threat is detected in Splunk AI, is the platform have response for it?

A. Yes - Through an integration with Cisco AI Defense, AI Agent Monitoring will detect and help prevent threats like privacy leakage or prompt injection. You will then have the ability to create a blocking policy. 

Q. In what version is it ready to use now?

A. It’s available now and the current version of that offering. We release continuously in our SAS environment, so we don't have Splunk version numbers for that. We do require that you use the latest version of the of the instrumentation libraries to be compatible with what we're releasing here on the application side.

Definitely check the Splunk Observability Cloud release notes on that and then, in the instructions on what versions of instrumentation and receivers. Technology is evolving very, very rapidly in this space and with every biweekly release, we're adding support for new platforms and technologies and frameworks. So, it's going to be something to always be aware of!

Q. So will the Splunk AI Assistant for SPL intergrated into search without having to download externally from the Splunkbase?

A. Yes, AI Assistant for SPL can now be integrated into search for a better search experience. However, it still requires an app download to allow us to interate and bring you the latest AI innovations quickly. We are exploring ways to make the AI Assistant for SPL a search default feature.