Bucheon Sangdong: research deliverables made with agents

The operator of UPMI Lab describes using agents to produce deliverables for the Bucheon Sangdong project. This case concerns the production of research and planning documents. It does not present construction or development outcomes as achievements of AI.

The retained materials include a research report, presentation-generation code, and a presentation file. The existence of a deliverable is different from evidence that its proposals have been implemented.

Turning research into a presentation takes more than copying paragraphs. A report needs a traceable argument; a presentation needs a clear question and a useful sequence of decisions.

Our review approach treats each claim, source, diagram, and explanation as a connected unit. A well-arranged slide still needs revision if it changes a number or omits an assumption from the underlying report.

The number of files produced does not fully describe productivity. Review time, revision effort, rebuilt slides, and claims without sufficient evidence all matter when assessing practical value.

This public introduction makes no numerical claim about time savings or productivity gains. A useful comparison starts with a defined task, source set, and completion standard, then records drafting and review time separately.

This introduction shares the production process and review approach. It does not reproduce detailed commercial terms or internal discussions from working documents.

A separate Bucheon project profile is available on the company website. That earlier profile and its planning visual are not presented as evidence of individual outputs or results from this agent-production case.