We've been building out a PDF viewer to help support various tasks in the Architectural, Engineering, and Construction space.
We use it primarily for steel estimation right now, but are exploring how it might be useful across other engineering disciplines!
From a tech perspective, the AI PDF viewers we've ran into use mostly PDF metadata/text, so they dont work well on engineering schematics which are mostly visual. So we put an agent in there that has tool abilities to see both text and image modalities, and its been much better for our tasks.
Hope its interesting to play with and open to ideas where the tech could be useful!