TECH TALKS

Native AI Intelligence: Splunk’s Hosted GenAI Foundation Models for Security and Time Series

Splunk Unveils Native GenAI Models for Security & Time Series in the Cloud

The value of Generative AI isn't in the hype—it’s in the specific architecture of the model, how the model was trained, and how seamlessly it integrates into existing investigation and observability pipelines.

Splunk is introducing the hosted AI models service, a new capability that allows you to pipe data natively into specialized LLMs for machine data and other open-source models, without provisioning GPUs or managing API access.

In this session, we will do a deep dive on how to access this service and the three model types now running directly within the Splunk Cloud Platform:

• For SOC Analysts (Foundation-sec-1.1-8b-instruct): We will explore this 8-billion parameter model, optimized specifically for security operations. Learn how it leverages training on cybersecurity frameworks to assist with threat intelligence correlation, alert contextualization, and interactive investigations using natural language.
• For SREs and Network Analysts (Cisco Deep Time Series Model): Discover the beta release of our Gen-AI, zero-shot forecasting model. We will demonstrate how to generate forecasts for predictive alerting and anomaly detection out of the box—with no model training or fine-tuning required on your data.
• For Complex Reasoning (GPT OSS 120B & 20B): See how we are utilizing a Mixture-of-Experts (MoE) architecture GPT model to balance speed and reasoning. We will show how the 120B parameter model handles complex root cause analysis and SPL generation, while the 20B model accelerates standard log analysis.

Stop building data science infrastructure and start engineering resilience. Register now to learn more and see a demo.

Our Speakers