The Journey to Autonomous Security - Building SentinelFlux

Why I’m Building SentinelFlux

I’ve spent a lot of time working with Wazuh and SOC tools. They are brilliant at telling you when something is wrong. They are the eyes of the network. But they aren’t the brain.

Currently, when a high-priority alert hits a SIEM, a human analyst has to scramble. They check the logs, they look for correlated patterns, and they manually run scripts to isolate the threat. In a world where AI-powered attacks happen in seconds, this human-speed response is becoming a bottleneck.

That’s why I’m starting a new journey: Building SentinelFlux.

The Core Concept

SentinelFlux is an AI-native orchestration layer that sits on top of existing security infrastructure. It’s designed to transform the SOC from “Detect and Notify” to “Investigate and Remediate.”

My goal is to use Agentic AI (specifically multi-agent workflows) to:

1. Self-Investigate: No more manual log hunting. The agent does it.
2. Understand Intent: Is that PowerShell script a threat or a maintenance task? Our AI will know.
3. Remediate at Scale: One-click (or zero-click) isolation of threats.

My Roadmap

Over the next few weeks, I’ll be sharing my progress in four phases:

• Phase 1: Setting the foundation with FastAPI and LangGraph.
• Phase 2: Building the “Investigator agent” to talk to Wazuh.
• Phase 3: Autonomous Response Orchestration.
• Phase 4: End-to-end simulation and “Survival Guide.”

Stay Tuned

I’ll be posting updates right here on my portfolio, sharing the code snippets, the architectural challenges, and the wins (and losses) of building a modern security tool from scratch.

If you’re in the AI or Security space, I’d love to hear your thoughts on this journey.