Lumi is a web app with AI-powered features to help you quickly read and understand arXiv papers. Features include:

• ✏️ AI-augmented annotations - summaries at multiple granularities
• 🔖 Smart highlights - highlight text + ask questions
• 🖼️ Explain figures - ask Lumi about images in the paper

Select text or click an image to ask Lumi questions:

Follow instructions in functions/README.md to install relevant dependencies and run local emulators.

cd frontend  # If navigating from top level
npm install  # Only run once

# Create an index.html file and (optionally) replace the placeholder
# analytics ID (see TODOs in example file) with your Google Analytics ID
cp index.example.html index.html

# Create a firebase_config.ts file and replace the placeholder.
cp firebase_config.example.ts firebase_config.ts

npm run start

Then, view the app at http://localhost:4201.

To view Storybook stories for Lumi:

npm run storybook

Then, view the stories at http://localhost:6006.

The import script in scripts/import_papers_local.py can be used to import a set of papers for local debugging.

The locally imported papers can be rendered in lumi_doc.stories.ts via Storybook.

To deploy the web app via App Engine, add an app.yaml configuration and set your Google Cloud project.

npm run deploy:prod

To deploy the Firebase cloud functions, see functions/README.md.

All software is licensed under the Apache License, Version 2.0 (Apache 2.0). You may not use this file except in compliance with the Apache 2.0 license. You may obtain a copy of the Apache 2.0 license at: https://www.apache.org/licenses/LICENSE-2.0.

Unless required by applicable law or agreed to in writing, all software and materials distributed here under the Apache 2.0 licenses are distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the licenses for the specific language governing permissions and limitations under those licenses.

This is not an official Google product.

Lumi is a research project under active development by a small team. If you have suggestions or feedback, feel free to submit an issue.

Copyright 2025 DeepMind Technologies Limited.

Lumi was designed and built by Ellen Jiang, Vivian Tsai, and Nada Hussein.

Special thanks to Andy Coenen, James Wexler, Tianchang He, Mahima Pushkarna, Michael Xieyang Liu, Alejandra Molina, Aaron Donsbach, Martin Wattenberg, Fernanda Viégas, Michael Terry, and Lucas Dixon for making this experiment possible!