The short version
Perplexity is an AI search engine that happens to take file uploads, not a chatbot that happens to search the web. That framing matters when you decide whether to upload a file or use a different tool. Where ChatGPT and Claude are great at talking to your documents, Perplexity is great at cross-referencing your documents against current web sources with citations for both.
If the question you're asking is "given what's in this industry report, how do those claims compare to what's been reported in the last month?" — Perplexity is the right tool. If the question is "given this 200-page document, help me draft a response" — Claude is the right tool. Use each for what it was built for.
What Perplexity does with an upload
When you attach a file in Perplexity:
- The file's text is extracted and added to the thread's context.
- When you ask a question, Perplexity decides whether to search the web, consult the uploaded file, or both — and cites whatever sources it used.
- Subsequent messages in the same thread keep access to the file. New threads start fresh.
- Spaces extend that pattern: files attached to a Space persist across every thread you start within it.
The blend of file content and web sources is the thing to understand. Perplexity will happily cite your PDF alongside a news article, and the answer pulls from both. That's the feature — and also the thing to watch, because when the two disagree, you want to make sure you notice.
Where Perplexity uploads shine
Competitive and market research
Upload a competitor's pricing sheet or product brief and ask Perplexity to compare it against what's publicly known about the space today. You get a synthesis that pulls from both your uploaded doc and fresh web sources, with citations you can walk through. Our post on competitive analysis with web scraping and AI covers the larger workflow.
Fact-checking against current sources
Upload a draft report or article. Ask "which claims in this document have been updated or contested by recent sources?" Perplexity will highlight specific claims and point to the web sources that update or challenge them. Useful for journalists, analysts, and anyone whose output has to stay current.
Research with live context
Upload a whitepaper or set of sources, then have a conversation that drifts between your uploaded material and the open web — always with citations. This is the use case Perplexity was built for, and where Spaces really earn their keep.
Spaces: the feature that pays for Pro
A Space in Perplexity is a workspace that holds:
- A set of uploaded reference files.
- A custom system prompt (the Space's "instructions").
- All the threads you've run inside it.
- Optional web source filtering (focus on specific domains or source types).
The recurring-topic pattern that makes Spaces useful: instead of a new conversation each time you open Perplexity, you have a home for each ongoing project. "My Q4 competitor landscape" Space holds competitor docs and an instruction to always compare against public news; every new thread inside it starts from that baseline.
Real workflows
- Sales research. Space per prospect. Uploaded: their latest 10-K, any public filings, recent press releases. Each new thread handles a specific question — the exec team, the latest earnings notes, what analysts are saying this week.
- Policy analyst. Space per regulation. Uploaded: the regulatory text, key industry comments, a summary of the rulemaking history. Instruction: always cite both the regulation and current news about enforcement or amendments.
- Journalist on a beat. Space per topic. Uploaded: the key academic papers and reports they keep returning to. Web search filtered to the specific publications they trust.
File preparation tips
| File type | Handles it well? | Tip |
|---|---|---|
| Clean PDF (born-digital) | Yes | Upload as-is up to the size cap. |
| Multi-column academic PDF | Unevenly | Convert to markdown first — citations come out cleaner. |
| Scanned document | Partial OCR | Use an AI-vision converter, review, then upload markdown. |
| Word doc | Yes | Fine as-is. Markdown is cleaner but usually unnecessary. |
| Excel / CSV | Supported | Small-to-medium sheets work; huge spreadsheets are better handled by Code Interpreter in ChatGPT. |
| Image | Yes | Good for screenshots, receipts, diagrams. Handwriting and dense tables: mixed. |
Where Perplexity isn't the right choice
- Long documents with no web component. Use Claude or Gemini. Perplexity's strength is search; an isolated long document doesn't play to it.
- Conversational drafting. If you want a long back-and-forth to work through a problem, Claude or ChatGPT are more conversational. Perplexity is built around shorter, answer-oriented exchanges.
- Heavily grounded research. When you need answers that come only from your sources (not from the open web), NotebookLM is a better fit — see uploading files to NotebookLM.
- Data analysis. For spreadsheet work, ChatGPT's Code Interpreter runs circles around any of the web-search-first tools.
Decision tree
Where to go next
If your pattern is "I have a bunch of sources I want to work across for weeks," NotebookLM is worth a look — uploading files to NotebookLM. If you're using Perplexity primarily for research that ends in a written deliverable, pair it with Claude for the drafting step: see uploading files to Claude.
And if the web pages you're citing are themselves messy, our web-to-markdown converter turns any URL into clean markdown you can paste directly into a Space.