AI-Driven Security

Use AI to support investigation and learning while keeping decisions and execution under human control.

Principles for AI assistance

Work from evidence

Explanations should refer to the supplied findings, logs, and configuration, identify uncertainty, and make assumptions visible.

Prefer explicit data choices

Local inference is part of the foundation. Any future external model integration should make the data destination and user choice clear.

Review before execution

Suggested commands and configuration changes should be understandable and reviewed before they are applied.

Documentation-aware guidance

A future retrieval workflow could consult selected technical documentation to support an explanation. Source selection, freshness, and answer quality would still need validation; retrieval does not eliminate incorrect answers.

Current stage

Ollama inference is verified in the lab. Retrieval, advisor agents, and automatic analysis of pipeline or runtime evidence remain proposals.

Explore the related DevSecOps vision