CurTask vs Elicit: from your own PDFs to a finished literature review

This page is for postgraduate students and researchers who are weighing up AI tools for a literature review and wondering where CurTask fits beside Elicit. Elicit is well known as an AI research assistant that searches a very large collection of papers, summarises them and helps with systematic reviews and data extraction. CurTask starts from a different place: the papers you already have, and the whole journey from reading them to writing your chapter.

At a glance

Your papers, not a search index

Most literature reviews are built from a folder of PDFs collected over months: open-access downloads, papers from your library's subscriptions, theses and older scanned articles, some of them in Malay. CurTask is built around that folder. You drop in PDFs, Word or OpenDocument files, and each paper is stored once in your Global Library. The duplicate check means the same paper is never stored or analysed twice, and the details are filled in from the DOI for you to edit.

Scanned pages are read in English and Malay on your own device, and the text is captured before the file is made smaller, so the AI always works from the whole document. A paper you only have the details of is saved and marked "No PDF": the AI analyses only papers that have their PDF, and you can add the file later. When you need more papers, you can search open research catalogues without leaving the app and add open-access PDFs in one click.

Evidence you can check against your own copy

Elicit shows sentence-level citations for its answers, and that is a real help. CurTask takes the evidence one step closer to your own work: every AI value in the literature matrix keeps the page number and a verbatim quote, and the quote is checked to exist on that page of your PDF before you see it. If no supporting passage is found, the cell says "No source found" instead of guessing.

You describe a column in plain words, such as "sample and context" or "theoretical framework", see the credit estimate first, then run it across your papers. Each value stays marked Suggested until you accept it, edit it or restore it. The researcher stays the author of the matrix, and the matrix exports to CSV whenever you need it.

Analysis that goes beyond the table

An extraction table is often where AI research tools stop. In CurTask it is one part of the Analysis lens. You can code passages as you read them, with a codebook whose colours stay easy to tell apart and coloured coding strips beside the text. Memos on coded passages stay yours and are never sent to the AI. Coding by hand is part of every plan. On paid plans, AI codes every paper of a project at the push of a button and sorts your codes into categories and themes; each suggestion stays Suggested until you accept or dismiss it.

Codes then flow into the synthesis map: a board and a mind map where you group codes into themes and mark which papers agree with, contradict or extend each other. When a theme is ready, you send it to Writing with its passages.

Writing in the same project

CurTask's Writing lens is a full manuscript space. You write in sections you can reorder, with binder, corkboard and outline views, and cite with @ as you type in any of eight styles, including APA, Chicago, Harvard and IEEE. Every citation can carry its page number and links back to the passage in the PDF through the evidence line, so you, your supervisor or an examiner can see exactly where a claim comes from.

Optional marks show the sentences that have no citation yet, and when you are done you compile to Word, PDF, Markdown or HTML.

The writing sidebar builds on your own accepted analysis rather than on a fresh search. It reads only the matrix values, coded passages, themes and links you have accepted, shows up to three cards for the paragraph you are on (contradictions first), labels each card with where it comes from ("From your matrix", "From your theme"), and shows its page and verified quote.

Available now in the Writing lens: Check my citations asks whether each paper you cited says what your sentence says, and Claim support marks the sentences a reader would expect a source for and looks for passages that back them in your own papers; on paid plans it also suggests published studies you don't have yet. Every passage from your papers is quoted and checked on its page, and your text is never rewritten.

A paper's analysis follows it everywhere

In CurTask, everything learned about a paper, its analyses and tags, travels with it into every project it joins. Reusing an analysis you already have costs 0 credits, so your second and third projects get cheaper as your library grows. Every AI run shows its cost before it starts, credits are never charged twice, and a job your balance cannot cover simply does not start.

Made for researchers in Malaysia

The interface is in English and Bahasa Melayu, and the AI writes in the language you choose for each project while quotes stay in their original language. Your papers, notes and writing are stored in Malaysia, private by default, and never used to train AI. Prices are in ringgit, with a free plan and a Student plan for university students.

When Elicit might suit you

If your main task is searching widely across a very large body of published research, getting quick summaries, or running a large systematic review with screening across thousands of records, Elicit is designed for exactly that and is worth a look on its own terms.

Using both

Many researchers search in one tool and work in another. You can use Elicit to find candidate papers, then download the PDFs you decide to keep and drag them into CurTask. The details are filled in from each DOI, and from there your matrix, coding, themes and chapter all live in one project, with every value traced back to a page in your own copy.

Start on CurTask's free plan with your own papers, no card needed, and see the evidence for yourself.