December 8th, 2023

🎁 Season Greetings! 🎄 As the end of the year rolls around, the Locus team is wrapping things up for some major improvements for PDFs and how the product looks and feels. This release adds all the goodness that html pages had to the PDFs, read on to get the details. There are also some fixes and pointers for feedback we have received from you, towards the end of the email.

All features for the webpage search are now available for PDFs

The search pop-up, recents, favorites, and summaries work the same as webpage searches. We still need to provide a separate PDF viewer, because Google says so. Here are some before and after screenshots:

Before, clunky search box inside the viewer

After, same pop-up and options available as regular webpage locus searches

It also includes the side panel features for Recents, Favorites, Summarization (both for document and sets of favorites, and available if you have a verified account).

Major Win! PDF file recognition.

This release also fixes an annoying issue where the extension didn’t recognize a page being a pdf because the source page did not have the pdf extension, or did not set the file type correctly.

Before, PDF not recognized. Boo.

Now, with some magic, we recognize a PDF, even if the page or file extension does not make it apparent.

Support for larger documents

We updated the Locus backend to be able to handle larger documents by implementing a cloud based worker system to process documents in parallel. From the extension perspective nothing looks different, but you’ll be able to load larger documents. Note that, for now, there’s still a 250k limit on number of words per doc, due to the amount of memory humongous documents take up.

Improvements in the Options Page and New Generative AI models for ChatGPT

OpenAI 3.5 Turbo is now available, and we also allow you to change the model for the indexing. Especially if you want to search by phrase instead of Q&A.