Chat with multiple PDFs and keep track of which file said what
Search+ lets you chat with multiple PDFs at once by putting them in one workspace and asking across the whole set. A single question can draw on several files, and the reply cites each document it used, so you can tell which file said what. When you need precision, narrow the same question to one document. Drag in up to 10 PDFs per upload batch and keep adding batches as the project grows.
Create an account without entering payment details.
Questions that only work across several PDFs
A multi-document chat earns its keep when no single file holds the answer. Here is what those questions look like and what Search+ returns.
A reply grouped by document, each point carrying a citation to its own file, so the three reports that mention delays are kept apart instead of blended.
A written comparison that cites the passage in each policy. You read both excerpts and judge the difference yourself.
A list of recurring points with citations to the transcripts where each one appears, which shows how widely a theme is shared.
Either the cited passages that conflict, or a statement that the files do not appear to address it. Both outcomes save you a full re-read.
The same question, re-scoped to a single document, with citations from that file alone.
What is a multi-PDF chat?
A multi-PDF chat is a conversation where the AI can draw on a whole collection of documents for each answer, rather than one file at a time. The hard part is not reading many files, it is keeping track of which file each statement came from once the answer mixes them. A useful tool therefore has to do two things together: retrieve relevant passages from across the collection, and label every part of the answer with its source. Search+ treats the workspace as the collection. Every indexed PDF in it is eligible, retrieval works by meaning across all of them, and each citation names the document and opens the excerpt. You also control the scope, so the same conversation can move between the full set and a single file.
How to chat with multiple PDFs in Search+
- Make one workspace per body of work
Name it after the project, the deal or the reading list. Everything you ask later is answered from the PDFs inside it.
- Upload the set in batches of up to 10
Drag up to 10 PDFs onto the upload area at a time. Each file shows its own status, and you can keep working while the rest finish processing.
- Ask across the workspace
Pose the question to the whole collection. The reply cites each document it relied on, so you can see how the answer is distributed across files.
- Narrow the scope when one file matters
Switch the question to a single document to get an answer drawn only from that PDF, then switch back to the full set.
Built for questions that span files
Questions that span several documents can be answered in a single query instead of one file at a time.
Each part of the answer points to the document and excerpt it came from, so mixed answers stay traceable.
Choose whether a question runs over the whole collection or one PDF.
Ask Search+ to compare two or more documents and say where they differ, with citations to each.
Ask for themes and repeated claims across the set and see which files each one appears in.
Drag and drop up to 10 PDFs per batch and watch each one move to indexed.
Workspace Context holds instructions such as preferred terminology or what counts as evidence, applied to every question in that workspace.
Limits worth knowing before you upload a stack
Search+ answers in written paragraphs and lists with citations, and those answers stay in the conversation. There is no downloadable table or file export.
A citation opens the supporting excerpt. It does not jump to a page inside a full document viewer, so keep the original file open if you want to browse it.
AI answers can be incomplete or inaccurate. Open the cited excerpt before you rely on an important point.
One workspace vs one chat per file
| What you need | One chat per PDF | Search+ multi-PDF workspace |
|---|---|---|
| An answer that combines three files | Ask three times, then stitch the answers together | Ask once; each part cites its file |
| Knowing which file a point came from | Rely on memory of which tab you were in | The citation names the document |
| Checking one file in isolation | Open that file's chat | Narrow the same conversation to one document |
| Adding new files next month | Start new chats | Upload into the same workspace and keep asking |
Questions about chatting with multiple PDFs
Can I chat with multiple PDFs for free?
Is there a GitHub project I can run myself?
How many PDFs can I upload at once?
Does the AI mix up which PDF said what?
Can I ask a question about just one of the PDFs?
Is this an AI tool or a search tool?
Can I get the comparison as a table or export it?
Will the set still be there next week?
Can the set include Word files and scans as well as PDFs?
Put the whole stack in one place and ask once
Upload your PDFs to a workspace and get answers that cite every file they use.
Start a workspace