update on politician lie detector

hi again! posted about InTruth last week and made some big changes so wanted to update! it now supports 16 languages including Spanish, Arabic, Hindi, Portuguese, and French!

just hit 11.4k users!!! always iterating on the pipeline, but it's the same idea:
 - text transcripted and chunks evaluated for check-worthy claim language
 - when a claim is detected, context around it is accumulated and sent to web search api
 - sources from web search call sent to claude for a verdict based on information the user can access in the UX, NOT training data

always looking for feedback and your concerns! thank you for so much support on the first post :-)