How it works
Context is a seven-step research workflow. Each step builds on the previous one. Analysis and drafts are built from the sources you provide, and their citations can link only to those sources.
A real sample

A literature review Context generated from open-access papers, unedited. On the sample page you can open its citations yourself. See the sample literature review
A workspace is the container for a single research project. Give it a title, write a brief description, and define your central research question. The research question matters — it keeps all AI analysis focused on what you actually need to know.
Be specific with your research question. Vague questions produce generic outputs. Specific questions produce useful ones.
Add your reading list to the workspace. Upload PDFs, paste URLs, enter a DOI, or create manual records for books, reports, or interview transcripts. Context extracts metadata automatically — title, author, year, journal, and abstract — and stores them as structured source records.
You can import sources incrementally. Start with a handful, run analysis, then add more. You do not need a complete list before you begin.
Select a source and click Run AI Analysis. Context reads the document's text, or the abstract if no file is attached (from a long document, it reads the beginning, three excerpts from the middle, and the end). It then extracts: a short summary, a detailed analysis, the main argument, methods, key findings, limitations, key claims, and notable passages. Each claim is labeled by evidence type and confidence level. Everything is stored in structured fields you can read, edit, and verify.
If you uploaded a PDF, Context extracts the full text first. Richer source material produces more detailed, accurate analysis.
After analysis, recurring themes and key concepts are identified across your sources automatically. You can view these from the workspace overview or the Concept Map tab. Each theme and concept links back to the sources it came from. Review the claims extracted from each source, add your own, or flag anything that needs verification.
The concept map view lets you see which ideas appear across multiple sources — useful for identifying the intellectual structure of a field.
Go to the Compare tab and select 2–8 sources. Context runs a structured comparison and returns: a synthesised overview, how main arguments relate, methodological differences, agreements, disagreements, conceptual overlaps, and gaps in the literature. The comparison is stored and can be exported as a report.
Use comparisons to find genuine contradictions and unexplored areas — not just summaries of what each source says, but where they diverge.
Open the Writing Studio. Choose which sources to include, select an output type (literature review, annotated bibliography, thematic synthesis, comparison table, research questions, chapter outline, or paper outline), and set tone and depth. Context generates a structured, cited draft from the sources you selected. Its citations link only to your sources, and an (Author, Year) citation that matches none of them is marked "Needs verification". Citations in a comparison table can't be opened and aren't marked, so check those yourself.
You can exclude sources you do not trust and regenerate. Use the verification panel to mark sections as verified, questionable, or needing revision.
Export any output as PDF, Word document (.doc), Markdown, or plain text. You can also export individual source analyses in the same formats, and comparison reports as structured documents. Citation lists export in APA, MLA, Chicago, BibTeX, or RIS. BibTeX and RIS import directly into Zotero, Mendeley, and EndNote.
Exports include the full works cited block and generation date. Treat them as structured starting points — not final submissions. Continue editing in your writing tool of choice.
Consistent throughout
Context writes analyses and drafts from the sources you added to your workspace, not from a web search, and the AI is instructed to cite only those sources.
A citation can link only to one of your sources; open it to see the source record and the passages extracted from it. An (Author, Year) citation that matches none of your sources is marked "Needs verification". Source records link back to the original file or URL where there is one.
You choose which sources to include in each output. You can edit any field. You can exclude a source, regenerate, and review what changed. Nothing is hidden or automated without visibility.
Common questions
Do I need to complete every step before I can use an output?
No. You can generate a literature review as soon as you have analysed at least one source. Running comparisons or exploring concept maps is optional — useful, but not required for basic output generation.
What if a source doesn't have a PDF?
Context works with metadata and abstracts too. Analysis will be less detailed than for a full document, but you will still get a summary, main argument, evidence type classification, and you can manually add claims and passages.
Can I edit the outputs after they are generated?
Yes. All outputs are plain text. You can edit them directly in Context, apply a refinement action (expand, shorten, restructure), or export them and continue editing in any external tool.
How many sources can I add?
The Free plan supports up to 20 sources in 1 workspace. Pro and Team plans remove these limits.
Can I rerun analysis on a source after adding more content?
Yes. Re-running analysis on a source overwrites the existing analysis with fresh results. If you have added notes, claims, or passages manually, those are preserved and re-run analysis adds to them.
The Free plan gives you 1 workspace and 20 sources — enough to work through the full workflow and see how Context fits your research process.