Best app for PhD students in 2026

Last updated May 2026

It was a study. Something about cortisol and decision-making under stress. Or maybe it was about cognitive load. You read it in your first year. You highlighted three paragraphs. It directly contradicts the argument you're writing right now, and if you don't cite it, your examiner will. You've spent two hours searching. It's in Zotero, or Google Drive, or that folder called "to read" that haunts you. You can describe what it said. You just can't find it.

This is the PhD tools problem. Not note-taking. Not flashcards. Not exam prep. Managing a multi-year research project across hundreds of papers, tracking evolving arguments, writing alongside advisors, and maintaining a literature base that grows every week for 3-7 years. The tools that work for undergrads don't work here. The problem is different.

Here are seven tools PhD students use. They're ordered by how much they help you build and manage a body of knowledge over years, not just store papers.

Quick comparison

Fabric

Fabric is the research library that reads your papers for you. Not literally. But every PDF you save is automatically extracted, indexed to the page, and connected to everything else in your library. Three years into your PhD, you can describe a study and find it. That's what matters.

Why it works for PhDs: The AI assistant understands your entire research library. "What have I read about cortisol and decision-making?" draws from PDFs, saved articles, lecture transcripts, and your own notes. Semantic search finds papers by what they're about, not by what you tagged them. In-document search finds the exact page in a PDF. Cross-platform search pulls from Google Drive, Notion, and Dropbox alongside your Fabric library.

Annotations on any content type. Notes editor with real-time co-editing for writing with your supervisor. Spatial canvas for mapping conceptual frameworks visually. Publishing with analytics for sharing work with your committee. Meeting transcription for supervisor sessions. Available on every device.

Your research compounds. Paper 400 connects to paper 12 because the Memory Engine mapped the relationship. You didn't create a link. You didn't tag anything. The system understood both papers and noticed the connection. That doesn't happen in any other tool on this list.

What it doesn't do: Citation generation. You still need Zotero (or Mendeley) for bibliography formatting. Fabric and Zotero are complementary: Zotero generates your citations, Fabric understands your library. Learn more about how researchers use Fabric. See also: best ways to organise research.

Zotero

Zotero is the reference manager. If you're doing a PhD in 2026, you're probably already using it. It handles the specific job of collecting, organising, and citing academic sources. It does this better than anything else.

Why it works for PhDs: 7,000+ citation styles. One-click citation insertion in Word, Google Docs, and LibreOffice. Browser connector saves papers from any academic database. PDF reader with annotation and highlighting. Group libraries for lab teams. Open-source. Free core app. The academic standard for a reason.

Where it stops: Zotero is a reference manager, not a knowledge base. It stores papers and generates citations. It doesn't understand what's in your papers. No AI. No semantic search (keyword and metadata only). No way to ask "what have I read about this topic?" across your library. Finding a paper means remembering the author, the year, or a keyword from the title. With 400 papers, that works. With 1,200, it doesn't. Zotero organises references. It doesn't help you think across them.

Obsidian

Obsidian is a local-first markdown editor with bidirectional links and a graph view. Many PhD students build their entire knowledge management system in it. See the full Fabric vs Obsidian comparison.

Why it works for PhDs: You read a paper, create a note summarising the key arguments, link it to related concepts, and over time the graph reveals the structure of your field. The Zotero integration plugin pulls in references. Dataview queries surface notes matching specific criteria. Full local ownership of your notes as markdown files. Free. Full offline. The PhD students who thrive in Obsidian are the ones who enjoy building the system.

Where it stops: You build everything. 5-10 hours of setup before it's useful. No AI (community plugins vary). PDFs are attachments, not indexed content. No semantic search. Every connection is one you created manually. The graph only shows links you made. If you forgot to link a paper to a concept, it's invisible. Many PhD students who start with Obsidian in year one have abandoned it by year two because the maintenance outpaced the research.

Heptabase

Heptabase is a visual knowledge tool. Cards on whiteboards. Mind maps. Bidirectional links. Built for people who think spatially. See the full Fabric vs Heptabase comparison.

Why it works for PhDs: Read a paper, create a card from your highlights, place it on a whiteboard alongside related concepts. Over time, the whiteboard reveals the shape of your argument. PDF annotation with highlight-to-card workflow. AI Tutor explains sources and researches topics with citations. Full offline. Good for building conceptual frameworks visually.

Where it stops: Every connection is manual. Limited content types. No semantic search. AI credits limited on Pro ($8.99/month), unlimited only on Premium ($17.99/month). No free plan. No collaboration beyond shared whiteboards. Heptabase works well for the conceptual mapping phase. It's less useful for the "I need to find that paper from 18 months ago" phase.

Readwise Reader

Readwise Reader handles the reading pipeline: articles, PDFs, ePubs, newsletters, YouTube transcripts. Highlights sync to a spaced repetition review system. See the full Fabric vs Readwise comparison.

Why it works for PhDs: The broadest reading tool in the category. PDF annotation with highlights that export to Obsidian, Notion, and Logseq. Ghostreader AI generates summaries and flashcards from highlights. Daily review emails resurface past highlights. For the specific workflow of read, highlight, review, export, Reader is the best pipeline available.

Where it stops: $9.99/month with no free plan. Reading only. No file storage beyond the reading library. No AI that understands your broader research. No semantic search across your collection. No spatial canvas. No collaboration. Reader is a reading pipeline, not a research home. It feeds your knowledge system. It isn't one.

Notion

Notion is the workspace many PhD students use for literature databases, project management, and writing. See the full Fabric vs Notion comparison.

Why it works for PhDs: Build a literature database with columns for author, year, methodology, key findings, and relevance to your argument. Multiple views (table, board, calendar) on the same data. Templates for reading logs, chapter outlines, and progress trackers. AI on Plus ($10/month) can summarise and answer questions about your Notion pages. Real-time collaboration with supervisors and lab members.

Where it stops: You have to build and maintain the system. PDFs are attachments, not indexed content. The AI only understands Notion pages. No semantic search. Notion's literature database requires you to manually enter metadata for every paper. At 400 papers, this is hours of data entry. At 800, it becomes a second job alongside the actual PhD. Many PhD students spend more time maintaining their Notion than writing their thesis.

Mendeley

Mendeley is Elsevier's reference manager. Free with 2GB cloud storage. PDF annotation and highlighting. Citation generation.

Why it works for PhDs: 2GB free cloud storage is more generous than Zotero's 300MB. PDF reader with annotation. Citation styles and Word plugin. Social features for discovering related research. Institutional access through many universities.

Where it stops: Declining. Elsevier acquired Mendeley in 2013 and development has slowed. The desktop app rebuild removed features. Users report sync issues. No AI. No semantic search. Limited plugin ecosystem compared to Zotero. Vendor lock-in with Elsevier. If you're starting a PhD in 2026, Zotero is the safer long-term bet. If you're already in Mendeley, it still works. It just isn't improving.

How to choose

If you want your research library to understand itself with AI that connects papers across your entire collection, semantic search that finds things by meaning, and no manual linking: Fabric. Pair with Zotero for citations.

If you need citation generation: Zotero. Non-negotiable for any PhD. Fabric doesn't generate citations. Zotero does. Use both.

If you want to build a visual knowledge graph of your field's conceptual landscape: Heptabase. Accept the manual work and the price.

If you want total local control over your notes with bidirectional links: Obsidian. Budget the setup time and the maintenance.

If you want the best reading and highlighting pipeline: Readwise Reader. It feeds your knowledge system. It doesn't replace it.

If you want a structured literature database with project management: Notion. Budget the data entry.

If you're already in Mendeley: Consider migrating to Zotero before your library gets larger. The switch gets harder every year.

The honest answer for most PhD students: You need two tools. One for citations (Zotero). One for understanding your research (Fabric, Obsidian, or Heptabase). No single tool does both. The question is whether you want the understanding to happen automatically (Fabric) or manually (Obsidian, Heptabase).

What most "PhD tools" articles miss

Most articles recommend the same stack: Zotero for references, Notion or Obsidian for notes, maybe Readwise for reading. This addresses the storage problem. It doesn't address the understanding problem.

A PhD is 3-7 years of accumulating knowledge. The papers you read in year one are relevant in year four. The argument that contradicts your thesis exists somewhere in your library. The connection between two subfields that nobody has noticed is waiting to be found in the gap between two papers you highlighted two years apart.

No amount of tagging, linking, or database maintenance surfaces those connections reliably. Manual systems are only as good as your memory and your discipline. Over 3-7 years, both erode. The tools that matter for PhDs are the ones that understand your library without you maintaining the connections, that find papers by meaning rather than by metadata, and that get more useful the more you add, not harder to navigate.

That's not a feature most tools optimise for. Most optimise for the first month. A PhD lasts 60.

FAQs

Do I need Zotero if I use Fabric?

Yes, for citation generation. Zotero handles bibliography formatting (APA, MLA, Chicago, 7,000+ styles) and in-document citation insertion. Fabric handles everything else: understanding your library, semantic search, AI Q&A across your research, annotations, collaboration, and publishing. They're complementary.

Can Fabric search inside my PDFs?

Yes. Fabric extracts and indexes PDF content automatically, making it searchable to the page. You can find the exact page that discusses a concept without opening the PDF. The AI can answer questions about the contents of your PDFs.

Which is best for literature reviews?

Fabric for finding and connecting sources across your library. Zotero for generating the bibliography. Obsidian or Heptabase for mapping the conceptual landscape of your field. Notion for tracking which papers you've read and their relevance to your argument. Most literature reviews benefit from two or three tools working together.

Should I use Notion or Obsidian for my PhD?

Both require building and maintaining a system. Notion is more accessible but cloud-dependent. Obsidian gives you local data ownership but a steeper learning curve. Neither understands your content the way Fabric does. The question is whether you enjoy system-building (choose one of those) or want the tool to do the understanding for you (choose Fabric).

Is Mendeley still worth using?

If you're already using it, it works. If you're starting fresh, Zotero is the better long-term choice: more active development, open-source, better plugin ecosystem, and no vendor lock-in with Elsevier.

Can I collaborate with my supervisor in Fabric?

Yes. Fabric supports real-time co-editing on documents, annotations on PDFs and other content, threaded comments, and shared drives. Your supervisor can review and comment on your work without needing to learn a complex tool.

How does Fabric handle 1,000+ papers?

Better the more you add. The Memory Engine maps relationships across your entire library. Semantic search scales with library size because it finds content by meaning, not by tag quality. A library of 1,000 papers is more useful in Fabric than a library of 100 because there are more connections to discover.