Outline arguments, compare sources, and turn chats into polished drafts.
Research
and
writing
with
a
calmer
academic
interface
Cortex brings drafting, source analysis, personalization, and DOCX export into one warm, distraction-free workspace — powered by Anthropic's Claude.
Anthropic's most capable model for nuanced reasoning, analysis, and long-form writing.
A distraction-free workspace that keeps you in flow from first prompt to final export.
Where research begins
Draft, analyze, and organize your work with AI in one focused place.
Frame your argument with a clear structure and confident academic tone.
Break sources into themes, findings, comparisons, and key debates.
Tighten clarity, flow, and tone across your existing writing.
Ask for a draft, upload a source, or describe what you need...
Features
Everything you need to go from question to finished draft
Cortex combines real-time AI conversation with document handling, personalization, and export — so your research workflow stays in one place.
Streaming responses
Read answers as they arrive and keep the pace of a real research session instead of waiting on full-page refreshes.
Upload real documents
Drop in PDFs, DOCX files, and notes directly so Cortex can reason over the material you are actually using.
Artifacts that stay usable
Turn chats into structured documents, reports, and thesis-ready drafts that are easy to export and keep working on.
Tuned to your field
Save your discipline, tools, and preferred tone once so every session starts closer to the way you actually write.
Security built in
Authentication, rate limiting, and secure storage stay visible as product strengths instead of hidden implementation details.
Export without friction
Download work as DOCX, JSON, or Markdown so nothing important gets trapped inside the product.
Core Stack
Built on infrastructure you can trust
How It Works
Move from idea to export in one flow
Start with a question, bring in your sources, shape the output, then take the result with you — all without leaving the conversation.
Ask or paste a draft
Start with a research question, a thesis claim, or a rough paragraph that needs structure.
Upload source material
Add PDFs, DOCX files, notes, or text so the model works from your actual material instead of generic assumptions.
Refine with context
Use follow-up prompts to compare arguments, rewrite sections, or tune tone for academic writing.
Export what matters
Save full sessions as JSON or Markdown and export polished outputs as DOCX when they are ready to leave the chat.
Summarize these sources and draft a thesis introduction.
- climate-policy-review.pdf
- notes-on-methodology.docx
- advisor-comments.txt
Use Cases
What people actually come here to do
Write introductions and sharpen arguments
Turn a topic and a few core ideas into a structured, credible starting point for academic writing.
Compare PDFs and extract key themes
Pull out findings, disagreements, and recurring patterns across uploaded research materials.
Rewrite drafts for clarity and tone
Keep the meaning, improve the structure, and make dense writing easier to defend or submit.
Ask technical questions with context
Use profile preferences and follow-up chat to get help that fits your tools, level, and working style.
FAQ
Short answers to the first questions people ask
What file types can I upload?
Cortex supports PDF, DOCX, TXT, Markdown, and CSV uploads for source analysis and file-aware conversations. Files up to 10 MB are extracted into the conversation automatically.
Is my data private?
Your chats, uploads, and profile are tied to your account and never sold or shared with third parties. Read the full Privacy Policy for the specifics on storage, retention, and deletion.
Are chats saved?
Yes. Sessions are stored in your account so you can return to them, search them, pin them, share them, or delete them later. Auto-delete is available in Settings if you prefer a clean history.
What are the usage limits?
Every account gets a rolling 200,000-token budget per five-hour window, shown live in Settings → Usage. Responses scale down automatically as you approach the limit so nothing gets cut off mid-sentence.
Can Cortex cite the sources I upload?
Yes. When you upload a document and ask a question about it, Cortex links its claims back to labeled source chunks so you can verify every reference against the original file.
Which AI model does Cortex use?
Cortex runs on Anthropic’s Claude family. Sonnet 4.6 is the default for everyday work; Opus 4.6 and Haiku 4.5 are available from the model selector for heavier reasoning or faster replies.
Can I export my work?
Yes. Sessions can be exported as JSON or Markdown for backup, and individual document outputs can be downloaded as DOCX directly from the chat.
Can I control what gets shared?
Yes. Shared sessions are opt-in and only visible through the generated share link while sharing is enabled. You can revoke a share link at any time from the session menu.