Workspace

Overview

The Workspace opens a full editing environment where you can refine every aspect of your scribe. Available to all tiers (Free, Pro, and Creator), it provides AI-assisted editing tools, research integration, and manual text editing in one place.

Accessing the Workspace

Open any of your scribes and click Workspace to enter the editing environment.

Workspace Layout

The Workspace has four main areas:

• Top bar — A single thin bar holds the scribe title (click to rename), the view-switcher tabs (Blurb / Scribe Hub / Read / Audioscribe / Workspace), and the three primary actions Tags, Export, and Share. The edition name sits as a tiny subtitle under the title — click it to rename. On phones the actions collapse into a "…" overflow menu.
• Sidebar (left) — Your scribe structure (chapters and sections). Click any item to select it for editing. Green checkmarks show completed sections. The sidebar auto-expands when you hover over it (after a brief ~150 ms delay so a quick cursor pass-by doesn't open it) and collapses when your mouse moves away (after ~350 ms so you have time to reach a section). Click the pin button to keep it open permanently — pinned sidebars stay open across section selections too, so you can quickly browse chapter-to-chapter without the sidebar snapping shut.
• Editor (centre) — Where you write your section content. WYSIWYG by default with a 15-button formatting toolbar at the top, the section title above it, and per-section instructions below. A <> Source View toggle is available for anyone who wants to drop down to raw Markdown.
• Right rail — Compact icon-first buttons for AI Write, Ask AI, Research, Verify, Upload, Cafe, Discard, and Save, grouped under tiny "AI Tools / Context / Document" labels. On phones the right rail collapses and the same actions appear in a floating action bar at the bottom of the screen.

What You Can Do

Edit Section Content

• Click on any section to edit its text directly
• You no longer need to know Markdown to write in the Workspace — the editor is WYSIWYG (what you see is what you get). Type normally; use the toolbar buttons or keyboard shortcuts to format
• Changes save automatically when you click away, or press Ctrl+S / Cmd+S

Formatting Toolbar

A 15-button formatting toolbar sits at the top of every section editor. From left to right:

Active formatting is highlighted on the toolbar — bold a word, and the B button turns primary-coloured.

Source View (advanced)

If you prefer to write in raw Markdown — or you've pasted in content with formatting that needs cleaning up — click the <> Source button in the section editor header. The editor switches to a plain <textarea> showing the underlying Markdown. Click Edit (the pencil icon) to switch back to the WYSIWYG view.

Source view is available to all tiers (Free, Pro, and Creator) — no tier gate. Your preference is remembered per-browser in local storage, so if you toggle on Source View it stays on across sections and reloads until you turn it off.

Keyboard Shortcuts Help

Click the ? button in the section editor header to open a small Keyboard Shortcuts popover. It lists every editor shortcut (Bold, Italic, Undo, Redo, Save, discard, indent/outdent). The popover replaces the old "Markdown help" cheat sheet — you don't need to memorise syntax any more, only the shortcuts you actually use.

Ask AI About Selection

Drag-select any text in the editor and a small floating menu appears next to it with an Ask AI about selection button. Click it to open the Ask AI panel pre-filled with the highlighted text — a fast way to ask the AI to explain, rewrite, or critique a specific paragraph without leaving the editor.

AI Write Cockpit

Click AI Write in the toolbar to open the generation cockpit — a two-column modal for full control over AI writing.

How the draft appears in the editor: AI Write generates the full draft on the server, then inserts the finished result into the editor in one step (rather than typing it out token-by-token). You'll see a brief loading state, then the formatted content appears all at once — ready to keep editing.

Left column — Writing controls:

• Write Mode — Choose how the AI handles your content:
  • Continue writing — Keeps your existing text, AI continues from where you left off
  • Improve existing — AI rewrites and enhances your current text (only shown when you have content)
  • Write fresh — AI writes new content from scratch, replacing anything existing
  • Write from research — AI writes the section based on your selected research notes
  • Expand from research — AI expands your existing text using your research notes (only shown when you have content)
• Instructions — A textarea where you tell the AI what to write. Type @ to autocomplete and reference saved research notes — selecting a note automatically checks it in the research panel on the right.
• Word Count Target — Quick presets (500, 1000, 1500, 2000 words) or type a custom target.

Right column — Context selection:

• Book Content — Check which parts of your scribe the AI should read as context: description, introduction, and individual chapters/sections. More context improves coherence but uses more tokens.
• Research Notes — A checklist of your saved research notes. Check notes to include them as context for the AI. Notes referenced via @ in your instructions are auto-checked.

Ask AI

Click Ask AI in the toolbar to open a question-and-answer panel. Ask the AI anything about your scribe, a section, or a general topic.

• Context modes — Control what information the AI receives:
  • No context — AI answers from general knowledge only
  • Book structure — AI sees your chapter and section titles
  • Section content — AI sees the current section's full text
  • Selected text — AI sees only text you have highlighted in the editor
• Research notes — Optionally include saved research notes as context
• Actions on responses — Copy the answer, Paste it directly into your section, or Save it as a new research note

Research Panel

Click Research in the toolbar to open the web search research panel. Use it to find information and save it for later use.

• Web Search — Type a query and the AI searches the web, returning an answer with cited sources
• Saved Research — View your saved research notes, click to open them in notebooks
• Context modes — Same as Ask AI (no context, book structure, section content, selected text)
• Save results — Save any research answer as a research note, which you can then reference in AI Write or per-section instructions

Document Upload

Click the Upload button in the toolbar to upload a document as research context. Supported file types:

• PDF (.pdf)
• Word (.docx)
• Plain text (.txt)
• Markdown (.md)

Maximum file size is 5 MB. Uploaded documents are saved as research notes that you can reference in AI Write, Ask AI, and per-section instructions.

Per-Section Instructions

Below each section editor, there is a collapsible instructions panel. Use it to give the AI persistent guidance for that specific section — for example, "Focus on practical examples" or "Keep this under 500 words."

• Click the panel header to expand or collapse it
• Type @ to autocomplete and reference saved research notes
• Instructions are saved per section and are automatically included when the AI generates content for that section

Bulk Generation

Generate content for multiple sections at once:

• Lightning mode — Generate all selected sections in parallel (fastest and cheapest)
• Pour Over mode — Generate sections with chapter-level context for better coherence

Structure Editing

• Add new chapters and sections
• Reorder chapters and sections
• Delete chapters or sections
• Rename chapter and section titles

@ References

Anywhere you see an instructions textarea in the Workspace (AI Write modal or per-section instructions), you can type @ to search and reference your saved research notes. When you select a note from the autocomplete dropdown:

1. A reference link is inserted into your instructions text
2. The note is automatically checked in the research notes context panel
3. The AI receives the full content of that note when generating

This lets you point the AI at specific research without manually copying text.

Token Cost Estimation

Before generating or rewriting content, the Workspace shows you:

• Estimated token cost for the operation
• Your current token balance

This helps you make informed decisions about which model to use and how much content to generate at once.

Model Selection

Switch between AI models within the Workspace. Each model shows:

• Name and provider
• Writing, Research, and Cost dot ratings (1-5)
• Tap the info icon for strengths and weaknesses, including a Citation behaviour (web search on) line for the curated favourites — useful when web search matters for your scribe

Picker structure (Creator only)

The model picker is a Creator feature. Free and Pro generate with the locked CoffeeScribe Model (our curated default) — no picker shown.

For Creator users, the picker has three tiers:

1. CoffeeScribe Model (recommended) — pinned at the top, pre-selected as the default.
2. Recommended models — ~12 curated alternatives we've personally verified for writing quality and citation behaviour.
3. All other models (advanced — at your own risk) — a collapsed disclosure that opens up the full OpenRouter catalogue (350+ models). Smaller models may produce noticeably lower-quality output, and some may fail to generate at all. Citation behaviour is unverified.

Text-completion writing models only — vision, image, audio, embedding, and code-specialised models are filtered out so you only see models suited to writing scribes.

If you select a model from the advanced disclosure, the picker shows an amber "Not in our recommended set — quality and citation behaviour are unverified, and the model may fail to generate" note under the model name. The model is still available — the warning is just a heads-up that we haven't personally verified it.

When a model is no longer available

Occasionally a provider retires a model endpoint. If you trigger a generation against a deactivated model, you'll see a clean error toast — "The selected model is no longer available — please pick a different model" — and no tokens are charged. Open the picker, choose another model, and re-run.

Imported scribes in the Workspace

Scribes created via the Upload your own flow (Scribe Conversion — see Importing your own books) work like any other scribe in the Workspace with two differences:

• Write Complete Scribe is hidden. The text already exists from the import — there's nothing for the AI to "complete." You can still rewrite individual sections, regenerate, or use AI Write per-section as normal.
• Too-big banner with "Convert to Scribe". If a scribe grows too large for the editor to handle smoothly (the section count exceeds the editor's safe limit), a banner appears at the top of the Workspace with a Convert to Scribe button. Clicking it pre-fills the import flow with the current scribe as the source — useful when you want to re-shape a sprawling work into a cleaner Mini-Scribe or AI Rewrite.

Tips

• Start with research — Use the Research panel to gather information before writing. Save findings as research notes, then reference them in AI Write for better, more grounded content.
• Use per-section instructions — Give each section specific guidance so the AI knows exactly what you want, even during bulk generation.
• You don't need to know Markdown — the WYSIWYG editor handles formatting. Click the toolbar buttons (or use Ctrl+B / Ctrl+I) and write normally. The export still produces the same clean PDF / EPUB / Word output.
• Use @ references — Instead of pasting research into instructions, type @ to reference notes. This keeps instructions clean and lets you reuse research across sections.
• Highlight + Ask AI — Drag-select a paragraph and use the floating Ask AI about selection button when you want a quick critique, rewrite, or explanation of a specific passage.
• Compare models — Try rewriting a single section with different models to find your preferred quality-cost balance.
• Save token credits — Use Lightning mode for independent sections; use Pour Over or Slow Brew when sections should build on earlier ones.
• Pin the sidebar — Click the pin icon to keep the sidebar open while you work. Unpin to let it auto-collapse when you move to the editor. A pinned sidebar stays open across all section selections.
• Drag-reorder sections within a chapter — Hover over a section in the sidebar and the drag handle appears on its left edge (always visible on touch screens). Grab and drop to reorder. Cross-chapter moves are coming in a later release — for now, drag-and-drop is intra-chapter only.
• Combine adjacent sections — Hover a section, click the ⋯ menu, choose "Combine with section above" or "Combine with section below". The two sections merge into one (the lower-numbered section's title is kept; the content is joined with a blank line). A 10-second Undo button appears in the toast if you change your mind. Any highlights and notes on the merged section are moved to the surviving section so nothing gets orphaned.

Section management — at a glance

Heads-up: existing audiobooks become stale after section edits

The audiobook is generated per chapter — each chapter's text becomes one saved TTS file. Reordering or combining sections, deleting a section, or editing text inside a chapter leaves that chapter's existing audio out of sync with the new text (the audio still plays — it just hasn't been re-narrated). Reordering whole chapters is free because each chapter's audio is unchanged — the player just plays them in the new order.

When you're happy with your edits, head to Audioscribe and regenerate the affected chapter(s) to bring the narration back in sync. See Audioscribe → When edits in the Workspace affect existing audio for the full table.