I’ve built and published Teklini Nexora, an evidence-grounded multi-agent platform for IT incident investigation.

The idea is straightforward: when an incident happens, engineers need more than an AI-generated answer. They need evidence, a traceable investigation process, validation, and clear boundaries around what the system is allowed to do.

Nexora combines:

• Deterministic, read-only diagnostic tools • Google ADK agent orchestration • Optional Gemini reasoning • Specialized investigation agents for network, systems, application, database, and security signals • Evidence-grounded findings with source and reliability metadata • Post-reasoning validation to challenge unsupported conclusions • Explicit uncertainty through NEEDS_MORE_EVIDENCE • Risk-tiered recommendations with a human approval boundary • FastAPI backend and React + TypeScript dashboard

The core design principle is:

Observe → Recommend → Approve → Execute

The current version intentionally stops at the approval boundary. It does not execute infrastructure changes or provide unrestricted shell access.

This was also an exercise in building AI systems where the model is not treated as the source of truth. Deterministic tools establish what was actually observed. The reasoning layer interprets that evidence, while a reviewer checks the resulting conclusions.

I’ve published the full project and documentation on GitHub:

https://github.com/bmcouma/...

Built as part of the Teklini Technologies engineering portfolio.

#AIEngineering #AIOps #SoftwareEngineering