🔄
Inline model swap for research writing
Draft the outline with a fast model like Gemini 3 Flash, then upgrade key sections to Claude Opus 4.8 for reasoning-heavy paragraphs, then finish with a lighter pass for polish. Everything stays in one document — no copy-paste between tabs. Each section carries a model-tag.
🎨
Six slide layouts for AI presentation
The renderer recognizes six layouts: title, content bullets, two-column comparison, big-number statistic, section divider, closing CTA. Prompt a 12-slide investor pitch and the AI picks the right layout per slide automatically. Export to Google Slides, PowerPoint, or Markdown.
✂️
Section regeneration without touching neighbors
Select a paragraph, hit regenerate, get three alternative drafts side by side — pick one, keep the rest of the document untouched. This is the core loop for tightening a long-form draft: you don't want the AI rewriting the whole thing just because paragraph 4 needs a stronger hook.
📊
Attachment-aware drafting from data
Drag a competitor's whitepaper, brand guidelines doc, or data screenshot into the session and the AI grounds the output in it. Especially useful for AI report generator workflows — feed it last quarter's numbers as a CSV and it produces the executive summary with the actual figures cited inline.
📤
Export to Word, Google Docs, PDF, PPTX
Every draft exports cleanly to Word (.docx), Google Docs, Markdown, PDF, and PowerPoint — headings, lists, callouts, links, footnotes, and slide layouts all preserved. No manual reformatting after export, no proprietary lock-in, no 'AI-detected' watermark on the output.
🌐
Draft in 90+ languages, translate inline
Write in Turkish, translate the whole document to English mid-draft with one command, or draft directly in Japanese with a French audience in mind. Model choice matters here: Gemini 3 Pro for Asian languages, Claude Opus 4.8 for nuanced European translation, GPT-5 for casual multilingual.