Multi-agent use case

Summary: I have used digital forensic tools/OSINT in the past such as Maltego and wwanted a tool I could integrate with AI. So I built my own Airgapped. This tool is the first iteration and will later be used to assist in high-risk controlled environments such as child protection agencies.

This is the current architecture and workflow.

Tools Used and function:

* Codex+Manus: Assistance in building the tool and incorporating logic. Bulk transfers of older method to current database. Data was collected by me and sorted into our database structure.

* Agents: Amending and adding bulk data to database.

* GPT+Manus: Verification and updates of data.

The final output:

Interface:

Inferences and patterns identified when AI (LLM+AGENTS) review data.

I add my own as well. Along with collaboration with AI to validate my understanding.

Evidence based Artifacts: All knowledge is sourced and tagged

These tie into a pattern identification graph so I can identify what may or may not be related.

Would love any feedback for improvements. Please remember, the next iteration is for child protection where I intend to airgap a localised LLM with training corpora. The main idea is to MINIMISE users from having to review images and identify patterns/locations to expedite rescue.

I want to add, this is also entirely self funded. I run a separate business to ensure I have funds for this and potential future hardware/licensing.