Perplexity Spaces vs Conversation History: How to Organise Your Research

Perplexity is primarily a research tool — its main advantage over general-purpose AI assistants is that it searches the web and provides cited, sourced answers in real time. As research use cases accumulate, the distinction between Perplexity's two history management systems matters more: your Library (all past threads) and Spaces (organised research workspaces).

Here's a clear breakdown of how each works, how to create effective Spaces, and how to use them together for serious research.

Perplexity Library: your full thread history

Your Perplexity Library is the base layer. Every thread you start — whether in a Space or just a standalone query — is stored in your Library, accessible from the Library link in the Perplexity sidebar.

What the Library contains:

• Every thread you've created, in reverse chronological order
• Thread titles (usually auto-generated from your first question)
• The full transcript of each thread, including Perplexity's sourced responses and the citations it referenced

What it's for:

• Returning to a past research thread
• Continuing a previous query with follow-up questions
• Reviewing the sources Perplexity cited in an earlier session
• Finding threads from a specific date range

What Library search covers: Perplexity's Library includes a search function that searches thread titles and related topics. It's more useful than many AI platforms' history search, but it still doesn't provide full-text search across the content within each thread — if you're looking for a specific data point, citation, or paragraph from a response weeks ago, the Library search won't reliably surface it.

The growing-sidebar problem: Active Perplexity users generate a lot of threads quickly. Because Perplexity is optimised for research queries (where you might ask ten related questions in sequence), threads accumulate faster than on general-purpose AI platforms. Without organisation, the Library becomes a long flat list within weeks.

Perplexity Spaces: organised research workspaces

Spaces are Perplexity's Pro-tier feature for structured, focused research. Think of a Space as a named research project with its own thread collection, shared files, custom instructions, and optional team access.

Spaces sit on top of the Library — threads within a Space still appear in your Library, but they're also accessible from the Space's dedicated view.

What Spaces add beyond Library organisation:

Custom instructions per Space. Each Space has its own instruction set that applies to all threads started within it. A "Legal research" Space might instruct: "Focus on UK jurisdiction unless I specify otherwise. Always cite the primary source. Flag when citing precedent that may have been overturned." A "Market intelligence" Space might specify: "Prefer sources from the last 12 months. Include publication dates with every source. Note when information is speculative vs documented."

File attachments. Upload documents — PDFs, reports, research papers, data files — directly to a Space. Perplexity can reference these uploaded files alongside its web search when answering questions within the Space. This is the key feature that makes Spaces genuinely different from just using folders: you can ground Perplexity's answers in your own documents, not just the open web.

Collaborative access. Invite team members to a Space. All members can start threads, add files, and access the Space's research. This transforms Spaces from a personal organisation tool into a shared research environment — particularly useful for research teams, consulting teams, or any situation where multiple people are researching the same domain.

Focused thread view. The Space shows only threads started within it, separate from the general Library. This makes it easy to see the full arc of research on a specific topic without the noise of everything else.

Creating a Perplexity Space: a practical walkthrough

Step 1: Create the Space. From the Perplexity sidebar, look for the Spaces option. Click to create a new Space. Give it a specific, descriptive name — "Competitor intelligence", "Academic literature — machine learning", "Due diligence — [company name]" — rather than something generic like "Research".

Step 2: Write effective custom instructions. This is the highest-leverage part of Space setup. Good Space instructions should include:

• Research scope: "Focus on B2B SaaS market data. Prioritise quantitative over qualitative sources."
• Geographic scope: "UK and US markets only. Note when findings differ between geographies."
• Source preferences: "Prefer peer-reviewed literature, government publications, and major industry reports over blog posts and opinion pieces."
• Time constraints: "Unless I specify otherwise, focus on sources from the past 24 months."
• Output format: "Structure answers with: 1) Direct answer, 2) Supporting evidence, 3) Caveats or contradicting evidence, 4) Sources."

Step 3: Upload relevant documents. For research Spaces that build on a fixed body of material, upload your starting documents:

• Market research reports or competitor analyses you already have
• Academic papers central to the topic
• Regulatory documents or legal texts
• Company filings or financial reports

These become part of Perplexity's reference set for every thread in the Space.

Step 4: Invite collaborators (if applicable). For team research, add collaborators via email. They'll get access to the Space's threads, files, and settings.

Step 5: Start your first thread. Begin researching within the Space. The custom instructions and uploaded files are automatically available.

Side-by-side comparison

The critical limitation: no cross-thread memory within Spaces

The most important thing to understand: Spaces do not give Perplexity memory of your past threads within that Space.

Starting a new thread inside a "Competitive intelligence" Space does not mean Perplexity has read your previous 30 threads in that Space. Each thread still starts fresh in terms of prior thread content.

What does persist across Space threads:

• The Space's custom instructions
• The files uploaded to the Space

What does not persist:

• The content of past threads
• Insights, findings, or summaries from earlier research sessions
• Any conclusions you reached in previous sessions

This is the same limitation as ChatGPT Projects and Claude Projects. Spaces provide a standing brief, not an accumulating shared memory.

Workarounds for cross-thread continuity

Since Perplexity doesn't carry thread content across sessions, users who need continuity within a research Space use a few approaches:

The running summary thread: Create one thread in the Space dedicated to capturing key findings. At the end of each significant research session, add your key conclusions, questions answered, and questions still open as a message in this summary thread. Over time, it becomes a living research log. Reference it at the start of any new session that needs continuity.

The reference document upload: Maintain a "research summary" document outside Perplexity (a Google Doc, a Notion page) that captures key findings from past threads. When the findings solidify, upload the updated document to the Space. Perplexity can then reference it in new threads.

The context paste: For sessions that directly build on a specific past thread, copy the key conclusions from that thread and paste them at the start of the new thread as context. Less elegant but precise.

How to use Spaces effectively by research type

Competitive intelligence: One Space per major competitor or per competitive category. Custom instructions focused on: pricing changes, hiring signals, product announcements, partnership moves, customer reviews. Uploaded files: competitor positioning documents, their public filings, analyst reports covering the competitive landscape.

Academic and scientific research: One Space per research domain or paper cluster. Custom instructions focused on: methodology quality, publication recency, study size, replication status. Uploaded files: key papers, systematic reviews, meta-analyses that anchor the research.

Due diligence: One Space per target company or deal. Custom instructions focused on: financial health signals, regulatory exposure, management track record, market position. Uploaded files: public filings, news archives, any proprietary information appropriate to share with AI tools.

Market research: One Space per market segment or geography. Custom instructions focused on: market size data quality, source credibility, recency. Uploaded files: existing market research reports, internal data summaries.

When to use Spaces vs just the Library

Use Spaces when:

• You have a defined research domain that will generate multiple related threads over weeks or months
• You have specific files or documents you want available as context in multiple sessions
• You're working with a team that needs shared access to research threads
• You want consistent research parameters applied automatically across all sessions in a domain

Stick with the Library when:

• You're doing one-off research questions
• You don't need to group or organise threads
• You're on the free tier and don't have Space access
• The research is exploratory and you're not sure yet what domain it belongs to

Searching across Perplexity history and Spaces

Neither Library search nor Spaces provide full-text search across the content of your threads. Finding a specific citation, a paragraph from a past response, or a data point from a research session from two months ago isn't possible through Perplexity's native interface.

LLMnesia indexes your Perplexity threads locally as you use the platform in Chrome. The index includes the full content of each thread — Perplexity's responses, citations, and your questions — and is full-text searchable. Since the index is local, your research doesn't leave your device when you search it.

If you use Perplexity for research and other platforms (Claude for drafting, ChatGPT for brainstorming), LLMnesia searches across all of them simultaneously — one query across your entire AI conversation history regardless of which platform you used.

Naming threads for better retrieval

Even without full-text search, thread naming significantly improves Library and Space navigability. After each thread, rename it with a specific, descriptive title:

• "Salesforce Q1 2026 earnings analysis" not "Salesforce research"
• "UK GDPR vs EU GDPR key differences" not "Privacy law question"
• "Transformer architecture — attention mechanism explainer" not "ML question"

Most platforms allow renaming from the thread's three-dot menu. It takes ten seconds and pays dividends every time you need to find that thread again.