Jacek points out the issue that approximately 95% of legislation is ignored because it exists as raw text that is difficult to read. To address this, he developed Lustra, which retrieves and processes bills in PDF or XML format from US and Polish APIs.
Product LaunchesJacekGemini 2.5 FlashLustra
Lustra removes political rhetoric from legal documents via LLM and implements importance ranking via citizen voting
This article is a translation. Read the Japanese original
For processing, Gemini 2.5 Flash on Vertex AI is used. With the temperature parameter set to 0, the system performs the removal of political rhetoric using strict JSON output.
The main feed is sorted by importance via a "shadow parliament" based on user voting, rather than editorial judgment. Draft bills submitted by citizens also undergo AI scoring and are displayed with the same visual priority as government bills.
The technical stack consists of Flutter, Firebase, and Google Cloud Run. The license is PolyForm Noncommercial, and a separate agreement is required for commercial use.
Currently, the service is operational for the US and Poland, with expansions to the EU, UK, and other regions in preparation. The developer is seeking collaborators to develop parliamentary data adapters for other countries.
The app can be accessed at lustra.news, and the code repository is available at github.com/fokdelafons/lustra.
Source: Show HN: Solving the ~95% legislative coverage gap using LLM's（HN 41pt・23 comments） (HN Search (backfill), 2025-12-16)