Prism is aura’s document workspace for heavyweight analysis. It exists so you can upload text-heavy files into a separate system, let aura process them in the background, and keep them ready for deeper retrieval-backed work later. This page explains why Prism is separate from the main chat surfaces, how the current documents workflow works, and what is coming next.

Why Prism exists

Prism exists because long-running chat and heavyweight document analysis are different jobs.

In Prism, you upload PDFs, DOCX files, and other supported formats into a dedicated document workspace. aura processes those files in the background and prepares them for later retrieval, comparison, summarization, and evidence-backed analysis.

The main chat modules do not offer traditional text-heavy file upload by design. Pushing large documents directly into long conversations can pollute the token window, degrade memory quality, reduce focus over time, and make costs climb faster than they need to. Separating document-heavy content into Prism keeps general aura chat lighter, more precise, and more economical, while still giving you a place for deeper document work when you need it.

How Prism works today

The current Prism experience is grounded in the documents module. The default path is:

1. Open Prism.
2. Click Upload.
3. Choose a supported file.
4. Wait while the file is processed in the background.
5. Search or filter the document list if needed.
6. Select the documents you want in scope.
7. Start a Prism chat from those selected documents.

This flow is intentional. Upload first, let ingestion finish, then start analysis once the content is ready.

Upload files into Prism

Prism currently supports these upload formats:

• PDF
• DOC
• DOCX
• PPT
• PPTX
• XLS
• XLSX
• TXT

The current size limit is 50 MB per file.

Uploads are explicit and manual. When you upload a file, aura starts an asynchronous ingestion workflow. The file is queued, processed in the background, and shown as processing while the workflow runs. Files are not instantly ready for document chat the moment they are uploaded.

Wait for background processing

After upload, aura prepares the document behind the scenes before it becomes useful for deeper analysis. That preparation step is what makes Prism practical for heavier use cases later.

This is why Prism feels different from a simple attachment picker:

• uploads continue processing in the background
• documents are prepared before chat depends on them
• the workspace is designed to keep prepared content ready for later retrieval-backed work

The result is a document system that stays ready for repeated analysis instead of forcing the chat model to re-ingest long text every time.

Search, filter, and manage documents

Once documents exist in Prism, the workspace gives you a document list instead of a one-off upload flow.

You can:

• search by file name
• filter by file type
• filter by source
• review files already added to the workspace
• download documents when needed
• use chat history from prior Prism conversations

This makes Prism useful as an ongoing document library, not just a temporary upload area.

Start a scoped Prism chat

Prism chats are created from selected uploaded documents. This is one of the most important differences between Prism and the general chat surfaces.

When you select documents from the table and start chat:

• the new Prism chat is scoped to those selected documents
• analysis stays anchored to the chosen set
• the system can perform summaries, comparisons, extraction, evidence gathering, and retrieval-backed reasoning against that scope

This is the workflow to use when you want complex document analysis without diluting a long-running general aura conversation.

Why this helps aura stay precise

Traditional file upload inside an always-on assistant thread can hurt the quality of long conversations. Large text payloads can fill the context window with document content that is only relevant for one part of the session.

Prism avoids that problem by separating heavy document content from the main chat modules. That separation helps in three ways:

• it reduces token pollution in long conversations
• it helps aura stay focused on the current task instead of carrying document bulk forward unnecessarily
• it keeps token costs more economical while still allowing advanced document analysis when you explicitly enter Prism

This is not a temporary workaround. It is the intended product boundary between lightweight conversational work and heavyweight document analysis.

What Prism is best for

Use Prism when the work is document-heavy and retrieval quality matters.

Good fits include:

• summarizing long files
• comparing multiple documents
• extracting evidence for a claim
• narrowing analysis to a selected set of files
• asking questions that should stay grounded in uploaded material

If the task is general discussion or lightweight reasoning, start from Home instead. If the task depends on uploaded documents and tight retrieval scope, use Prism.

Upcoming capabilities

Prism is also the foundation for a broader analysis pipeline that is still being built.

Coming soon:

• multimodal retrieval pipelines for richer document understanding
• stronger support for Excel, CSV, and other structured data workflows
• deeper structured-data analytics for complex pattern extraction
• data-science style analysis workflows
• generated charts and graphs
• generated reports from uploaded analytical material

These capabilities are upcoming, not current behavior. The goal is to make Prism the place where aura handles not only long-form text documents, but also richer analytical and structured-data workloads.

Good to know

Prism is separate from Connections. You do not configure Prism as a connector. It is the dedicated workspace for uploaded document analysis.

Next: Home or History