Working with PDFs

Upload a PDF to your project and the agent can read it — extract text, answer questions about the contents, summarize sections, or pull data into a sheet. No manual copy-pasting required.

Upload a PDF

The quickest way: drop it into the chat and tell the agent what you need.

Summarize this PDF and list the key takeaways.

Drag the file into the chat input area, add your prompt, and send. The agent picks up the PDF immediately.

You can also click the paperclip icon (📎) in the chat input to browse and select a file. To upload without starting a conversation, click + in the toolbar and select Upload file. The PDF appears in the Files panel, accessible from the paperclip icon (📎) in the right sidebar.

What the Agent Can Do

Once a PDF is in your project, the agent reads the full document — text, tables, layout. Here's what you can ask.

Extract Text and Data

Extract the data table from page 3 and put it in a sheet.

The agent pulls text from any page, identifies tables, and can structure data into a workbook you can filter, sort, and query.

Summarize and Analyze

What are the main findings in this report?

The agent summarizes at whatever level you need — a one-paragraph overview, bulleted key points, or a section-by-section breakdown.

Answer Specific Questions

What's the termination clause in this contract?

Point the agent at a specific detail and it finds it. Works well for contracts, financial reports, and research papers where you need one piece of information without reading the whole thing.

Pull Data Into Sheets

Create a sheet from the financial data in this PDF — revenue by quarter.

When a PDF contains tables or lists, the agent extracts the data into a workbook sheet. From there you can run formulas, build charts, or combine it with other data.

Highlight Sections

Highlight all the liability clauses in this contract.

The agent finds matching text in the PDF and highlights those regions in the PDF viewer — words, paragraphs, tables, or entire sections. Highlights include a label showing the matched text. Ask for specific regions on particular pages to narrow results — the agent can search within a page range or focus on specific element types like tables or paragraphs.

Useful for contract review (liability and termination clauses), financial analysis (key figures and totals), and research papers (findings and methodology sections).

Tips for Best Results

Be specific with your prompts. "Summarize this PDF" works, but "Summarize the risk factors section in three bullet points" works better.

Reference page numbers when you can. If you know the data is on page 3, say so. It helps the agent focus.

Use chat upload for immediate work. Drop the PDF in the chat with your request for the fastest path from file to results.

What to Expect

Obvious processes the full text of your PDF on upload. A few things to know:

• Scanned PDFs (image-only, no selectable text) are harder to work with. The agent uses visual analysis, but PDFs with selectable text give the most reliable results.

• Large PDFs (100+ pages) may take a moment longer to process. The agent still reads the full document.

• Password-protected PDFs need to be unlocked before uploading. Obvious can't open encrypted files.

Next Steps

• Uploading & Managing Files — getting files into your project and organizing them.

• Files & Media Overview — how all file types work in Obvious.