Vertical AI for Scientists

A specialized AI research assistant that can access paywalled academic articles using user credentials, enabling researchers to search and analyze full-text content from academic journals.

Key Points:

• Problem:

  • Academic research is locked behind paywalls (journals like Elsevier)
  • Existing AI tools can only search abstracts or open web content
  • Researchers can't easily access or search full-text articles even with credentials

• Solution:

  • Create a version of Deep Research that allows users to enter their paywall credentials
  • AI would log in on behalf of the user to access full-text articles
  • Index the full text of journals with user credentials

• Market Context:

  • Academic publishing is a controversial industry with high profit margins
  • Researchers and peer reviewers don't get paid, but publishers charge "an arm and a leg"
  • Existing competitors (Elicit, Syspace) only allow searching abstracts or uploading individual PDFs

• Technical Approach:

  • Similar to Deep Research but with credential storage and login capabilities
  • Not terribly difficult to build the base functionality (similar to GenSpark)
  • The credential storage and login functionality would be "a little bit more tricky"

• Broader Opportunity:

  • "The more private data you can get access to, the more useful these agents become"
  • Could potentially expand beyond academic research to other auth/paywalled content