TreatyHopper was a tool that could give multinational corporations a starting point for lowering their effective tax burden on international transactions by routing profits through intermediary countries with favorable tax treaties. AI agents gathered the relevant treaty rates to populate a directed graph modeling the global treaty network. Given a source and destination jurisdiction, the tool then applied Dijkstra's algorithm to find a path that minimized the total withholding tax along the route. The result was intended as a starting point for further tax planning, not a fully optimized end-to-end solution.

The project has been archived. The hosted demo incurred ongoing infrastructure costs disproportionate to its usage, so I have taken it offline and preserved it here as a visual record. The methodology is described in detail in my blog post.

ARCHIVED 1 / 4

Figure 1: Landing page introducing treaty shopping, with the expanded “Read before continuing” advisory panel outlining caveats about tool accuracy and user responsibility.

← Previous

Next →