CGIP AI Assistant
Answers from the orders themselves.
For staff sections preparing for a Commanding General's Inspection, who build their checklists by cross-referencing orders line by line. Built by RIG's founder in Marine Corps service.
Who it is for
Preparing for a Commanding General's Inspection Program (CGIP) evaluation means reconciling a unit's programs against a large, shifting body of official directives. Staff sections build checklists by hand, cross-referencing MCOs and NAVMCs line by line. It is slow, repetitive work, and a missed update or stale citation carries real consequences for the unit's inspection results.
What changed in their work
The assistant generates functional-area checklists covering roughly 84 inspection areas, and every item cites the governing directive text. Staff can check each requirement against the order itself instead of trusting a paraphrase, and can filter by core, enhanced-core, and non-core requirements. It was delivered working end to end and served as the live demonstration at the 2026 Marine Corps AI Workshop.
How adoption was designed
- Trust through citations. People adopt AI output they can verify. Every item points to the order it comes from, so a staff officer can defend it in front of an inspector.
- Built for the network units actually use. Restricted cloud access, blocked serverless runtimes in AFPIMS, and workstations without administrator rights shaped the design, so the tool could be used where the work happens.
- Organized the way inspections are. Checklists are structured by functional area, so the tool extends to new areas without a rebuild.
How it is built
A React single-page application (Material-UI, installable on mobile) with language-model processing through an API and a serverless execution layer, on an open-source stack. The network constraints it ran into now shape RIG's local-first tools that need no backend.