Open-source experiments where operations meet the algorithm.
This is where I build and publish small, sharp tools for the parts of supply chain and operations that software still walks past - the physical work, the tribal knowledge, the decisions nobody ever wrote down. Each project is MIT-licensed, built in public, and small enough to read in one sitting.
The lab, with every project and its status, lives at automati.qa. The build logs and the thinking behind each one are on alxsidr.io.
| Project | What it does | Status |
|---|---|---|
| oodaa | A small, readable self-improving agent loop - Observe, Orient, Decide, Act, and the second A, Adjust | live |
| risk-navigator | Multi-agent operational risk monitoring across freight, fuel, labour, weather, and geopolitics, turned into briefings and a live dashboard | live |
| flowtwin | Watch an operational process once, get editable process maps and runbooks back | work in progress |
| orchestrator | The execution layer - turn signals into decisions, policy, and a traceable record | work in progress |
| calibri | Agentic coordination of sample lifecycle management for agri-food and soft commodities | pipeline |
| synthax | A synthetic assistant for supply chain - voice-first and context-aware | pipeline |
Most agentic AI assumes a level of digital maturity that real operations do not have. The process lives in someone's head, the exception gets handled by instinct, and nothing downstream can be automated because nothing upstream is described. The lab works the other side of that gap: capture what actually happens, make it legible, and only then let an algorithm act on it.
Everything here is open. Fork it, break it, send it back better.
- Lab: automati.qa
- Build logs: alxsidr.io
- LinkedIn: Aleks Sidorecs