Excited to see how the recursive agentic loops improve model performance. Sounds promising!

Has anyone tested Gemini 3.8 Flash Cyber in a real-world cybersecurity scenario yet?

The term 'agentic workflows' seems a bit vague. Would love more concrete examples of its application.

Interesting read on Gemini 3.8 here: https://blog.google/innovation-and-ai/models-and-research/gemini-models/3-8-flash-and-3-8-flash-cyber