arjunworks All products Open the app

Research Observatory · A product by Arjunworks

Read the answer.
Inspect the evidence.

Ask a question about a public research paper, see the passages used to answer it, and open the cited PDF page. Research Observatory brings the model answer and its source material into one workspace.

Open the app Request access

Live research preview. A passcode is required. Access is approved manually.

From a question to the source page

  1. Choose a paper. Start with one of the bundled papers, or add a public text PDF up to 4 MB.
  2. Ask a specific question. Select a paper or search the shared library. The model runs only when you press Ask.
  3. Follow the evidence. Select an inline citation, a source-map point or a retrieved passage to inspect the physical PDF page.

For a first question, choose Attention Is All You Need and ask why scaled dot-product attention divides by the square root of d_k.

What to expect from this preview

  • Real generation, visible sources. Answers come from a connected language model. Citations identify returned passages; they do not guarantee that every claim is correct.
  • Public papers only. This is a shared library, visible to other approved visitors. It is not a private document workspace.
  • Saved paper library. Uploaded papers remain in the shared library after backend restarts. Keep your original files: this preview is not a backup service. You may need to select your original PDF again to view its pages in a later browser session.
  • A limited shared allowance. The app displays the model-call allowance. Access does not include unlimited usage.

Try the research preview

Request access by email

Tell Arjun which public papers you want to explore and what you would like to test. Requests are reviewed manually; a passcode is shared privately when access is available.

Open an email draft

This opens your email app. Nothing is submitted automatically, and no account is created. Send the message from your email app when ready.

Or write to ponnagantiarjun644@gmail.com. Please do not include confidential papers or access credentials.

Also from Arjunworks: Evidence Audit

Our main product development focus is Evidence Audit: an early-stage prototype for examining AI-generated claims against retrieved sources, one claim at a time.

Explore Evidence Audit and its current evidence