Show HN: I built research agents because GPT cant accurately hilite txt in files

app.ubik.studio

2 points by ieuanking 13 hours ago

Hi HN,

We built Ubik because no available Chatbots, AI research platforms, or AI PDF analysis tools could highlight text down to the line level in the file.

Why is this important?

Researchers, academics, scientists, and students lack the tools to help produce usable research while elevating workflows without over-automating the steps between A-Z. They also need to trust the accuracy of output/analysis generated by the agents & models, and confidently gather quotes, sources, and citations for scholarly papers.

Key Features of Ubik:

Search academic databases and convert open-access papers into AI-ready docs. Use enhanced OCR + PDF analysis tools. Make AI notes in PDFs that highlight text down to the line level. Reference files when prompting using the @ symbol (like Cursor), which minimizes hallucination and improves output. Generate high-level text with citations and use over 20+ models.

We are working on a custom EVAL suite and will publish our findings soon.

Please reach out with any comments, questions, or critiques.

ieuanking 13 hours ago

Forgot to mention:

Our academic search uses ArXiv and Semantic Scholar, so when using the academic agent, you can convert open-access papers into AI-ready docs directly from the databases. <33

  • ieuanking 12 hours ago

    Also, you do not need to sign up to use Ubik - I am fixing that copy right now :/