Everything you can do
with Innogath
Innogath is a research workspace that turns questions, files, links, and follow-up conversations into structured reports, visual data blocks, citations, and branching pages. This guide walks through the current workflow — from creating a project to using Fast, Thinking, and Deep Research modes; reading source evidence; branching from reports or diagrams; writing in the Notebook; and exporting your work.
Getting Started
Create an account, open your first project, and learn the workspace layout.
Create an account
Head to app.innogath.com/register and sign up with email or Google. A verification link is sent to your inbox — click it to activate your account. You can then sign in at app.innogath.com/login.
Every new account starts on the Free plan with monthly credits that cover your first Deep Research report and a couple dozen quick chat messages. No credit card required.
Your first project
Once signed in, create a project from the left sidebar:
- Click + New Project in the sidebar.
- Type your research question in the chat input — the more specific, the better.
- Use the + menu if you need to attach files, images, or links before sending.
- Either press
Enter(Auto mode decides the strategy) or click Deep Research for a full report. - Watch the Activity Stream as the system plans, reads web or file sources, and writes the result.
- When it finishes, the result appears as a report page. Depending on the task, it may also include visual data blocks, tables, references, and an auto-generated canvas.
Interface overview
The workspace is split into three zones:
- Sidebar — projects, recent pages, Notebook, and Settings.
- Book Pane (left) — your research report, displayed as numbered chapters with inline citations.
- Canvas Pane (right) — the auto-generated knowledge diagram plus the chat box for follow-ups.
Above both panes is the toolbar: panel toggles, Refresh Diagram, Edit, Pages Overview, and Export as.
Panel switching
You can hide or show either pane at any time:
- Book only — collapse the canvas for distraction-free reading; the report expands to a comfortable two-column layout.
- Canvas only — collapse the book to focus on the diagram and chat.
- Split view — the default: both panes side by side, resizable via the divider in the middle.
Panel state is remembered per project, so you can use different layouts for different research topics.
Deep Research
The core feature — full structured reports backed by web sources, uploaded files, or earlier research.
What is Deep Research
Deep Research is a specialised workflow that plans, reads, and writes a full report with inspectable evidence. Runtime depends on the task: a focused file analysis may finish quickly, while a broad web research run can take much longer. Unlike a single-shot chat reply, it actively:
- Plans the work — breaks the request into the sources, files, or subtopics that need attention.
- Reads the evidence — web pages, uploaded documents, CSVs, screenshots, or earlier project reports depending on the prompt.
- Cross-checks claims — separates supported facts from inference and flags conflicts or weak evidence.
- Writes the report — organised into sections with citations, references, and when useful, visual data blocks.
- Creates follow-up structure — report pages can branch into new pages, chats, or canvas nodes.
How to trigger it
There are two ways to start a Deep Research run:
- Click the Deep Research button next to the chat send button when you explicitly want a report-style answer.
- Use Auto mode (the default). Auto keeps simple requests in chat and sends broader research tasks to the deeper workflow.
You can also trigger Deep Research from inside a chat by clicking Deep Research on any existing thread — the current conversation becomes context for the new report.
Activity Stream
While the AI works, the Activity Stream shows live progress as a compact timeline. Activity rows are expandable so the page stays readable during long runs:
- Earlier research — shows when the answer is using parent pages or previous reports as context.
- Web search — shows search planning, URLs returned, and source-read progress.
- File reading — shows attached-file intake for PDF, image, CSV, or document analysis.
- Writing — shows report drafting, synthesis, and finalisation status.
Use the stream as an audit trail: it tells you whether an answer used web search, earlier research, uploaded files, or a mix of sources.
Report structure
Reports arrive as a structured document in the Book Pane:
- Numbered chapters — clearly demarcated sections with a consistent heading style.
- Serif typography — long-form reading comfort, with key findings in bold.
- Inline citation tags — small badges such as Source Name near factual claims.
- Visual data blocks — tables, charts, comparison matrices, and expandable data views when the task is data-heavy.
- References section — a consolidated list of web and file evidence used by the report.
- Two-column view — when the canvas is hidden, the report expands for more comfortable reading.
Citations & verification
Research outputs are designed to be inspectable rather than treated as a black box:
- Click web citations to open or inspect the original source.
- Click internal file references to open an evidence modal showing the cited file and evidence unit.
- Open expanded tables when a report includes large structured data.
- If evidence is incomplete, the report should say what is supported, what is inferred, and what still needs checking.
Audit the evidence. Citations and file references make claims easier to inspect, but important decisions should still be checked against the underlying source.
Chat Modes
Four distinct modes, each tuned for a different task. Pick manually or let Auto decide.
Change the mode using the dropdown next to the send button. You can set a default in Settings → Default chat mode.
| Mode | Best for | Searches web? | Speed |
|---|---|---|---|
| Auto | Most questions — the AI picks the right strategy | Sometimes | Varies |
| Fast | Quick answers, brainstorming, simple follow-ups | No | Seconds |
| Thinking | Slower analysis, source-aware follow-ups, file or image reasoning | Sometimes | Varies |
| Deep Research | Full structured report with citations, references, or visual data blocks | Often | Several minutes+ |
Auto
The smart default. Auto reads your question and routes it: short prompts stay in Fast chat, deeper analysis goes to Thinking, and broad report requests can become Deep Research. Use Auto when you want the product to choose the lightest mode that can still answer the request.
Fast
A direct chat reply for quick answers, brainstorming, and simple follow-ups. Fast mode is best when you do not need a full evidence-gathering workflow.
Thinking
For questions that need slower analysis without turning into a full report. Thinking mode can use conversation history, parent-page context, uploaded material, and lightweight web search when needed. It shows user-facing activity such as source use or drafting status, not a raw private reasoning trace.
Deep Research
The full workflow described in Chapter 2. Use it when you want a durable research artifact: a report page with citations, references, tables, diagrams, file evidence, or a branchable canvas.
Attachments & links
You can enrich any chat message with attachments from the + menu:
- Files — upload PDFs, CSVs, documents, or spreadsheets for source-grounded analysis.
- Images — upload diagrams, screenshots, or photos for visual analysis.
- Links — paste or add URLs so Innogath can inspect external pages when the selected mode allows it.
Uploaded files can be used in chat answers, internal-source analysis, or Deep Research reports. If you need a report to rely only on uploaded files, say that directly in the prompt.
Reply with quote
One of Innogath’s most powerful features. Select any text inside an existing report, then click Reply in the floating toolbar:
- The selected text becomes a visible quote in the new chat context.
- A new branch is created under the source section (see Chapter 5).
- The parent report’s research memory is inherited — the AI won’t re-verify facts it has already established.
This is how you drill down from a high-level finding into a specific sub-topic without losing context.
Canvas & Diagrams
Every report gets an auto-generated visual knowledge map you can refresh, edit, and export.
Auto-generated canvas
The moment a Deep Research run finishes, Innogath picks the best diagram type for your topic and renders it on the Canvas Pane. The diagram:
- Reflects the structure and relationships described in your report.
- Uses colour-coded nodes for different sections (blue for concepts, green for actions, amber for warnings).
- Supports smooth zooming, panning, and fit-to-view.
Refresh diagram
Click the Refresh button (the circular arrow icon in the toolbar) to regenerate the diagram from scratch. Use this when:
- The current diagram doesn’t match your mental model.
- You’ve added new information to the report and want the canvas updated.
- You want to try a different layout for the same content.
Edit mode
Click the Edit button (pencil icon) to enter edit mode. You get an Excalidraw-style editor with:
- Shape tools: rectangle, diamond, ellipse, line, arrow, freehand draw.
- Text labels with custom fonts, sizes, and colours.
- Move, resize, rotate, and delete any element.
- Stroke styles: solid, dashed, dotted. Fill patterns: solid, hatched, cross-hatched.
- Group, align, and distribute multiple elements.
Click the Edit button again (or press Escape) to exit edit mode. Changes save automatically.
Click a node to chat
In view mode (not editing), click any node in the diagram to open a focused chat context for that topic. From there you can:
- Ask follow-up questions in Fast, Thinking, or Auto mode.
- Trigger a new Deep Research run — the result becomes a new page branching off the current one.
- Select text from the node’s content and send it to the Notebook.
Visual formats
Innogath picks from a visual library based on your topic. The canvas renderer auto-selects a useful structure, and reports can also include inline visual data blocks such as tables, charts, and comparison matrices:
Export canvas as PNG
In the canvas toolbar, click Export → PNG to download the current diagram as a high-resolution image. The export respects your current zoom level and theme, and includes:
- The full diagram at 2× pixel density (retina-quality).
- A transparent or white background (your choice).
- Optional Innogath watermark in the bottom-right corner.
Pages & Branching
Your research grows as a tree, not a linear thread. Every sub-topic is a new page with inherited memory.
What is a branch
A branch is a child page that spawns from any point in an existing report. It has its own chat, report, and canvas, but inherits its parent’s research memory. Branching is how Innogath stays organised as your exploration grows.
Each branch node in the Pages Overview tree corresponds to one Page. A project can have an arbitrary depth of branches: great-grandparent → grandparent → parent → current.
Branch from a report selection
The most common way to branch:
- Select any text in the Book Pane — a paragraph, a citation, a single sentence.
- Click Reply in the floating toolbar that appears.
- Ask a follow-up question, or trigger Deep Research for a full sub-report.
- A new child page is created under the source section, with the quoted text as context.
Branch from a canvas node
You can also branch directly from the visual canvas:
- Click any node in the diagram to open a focused chat context.
- Ask a question or click Deep Research.
- If you run Deep Research, the result becomes a new child page linked to that node.
This is particularly useful when you’re visually exploring a topic — you can see what deserves deeper investigation and drill in with one click.
Pages Overview tree
Click the Pages Overview button in the toolbar to see the entire project as a tree:
- Each node is a Page (a report + its canvas).
- Lines show parent → child relationships.
- Node labels show the page title and a small preview.
- Click any node to jump directly to that page.
- Breadcrumbs at the top of every page show the full path:
Root → Main Report → Sub-topic A → Detail B.
Parent memory
When you branch, the child page inherits a summary of every ancestor’s research. This means:
- The AI knows what you’ve already covered and won’t repeat information.
- Follow-up questions get answers in context, not from scratch.
- The memory chain traverses multiple levels — grandparent insights feed into great-grandchild reasoning.
- Citations from parent reports are visible in child chats so you can reference established facts.
Pro tip: for wide topics, start with one high-level Deep Research at the root, then branch narrowly into specific sub-questions. You’ll spend fewer credits and get better answers than running ten separate research runs.
Notebook
A rich-text scratchpad bound to your project. Capture insights, draft writing, collect quotes.
Opening the notebook
Click the Notebook button in the left sidebar (the notebook icon). The Notebook opens as a full-height panel that overlays the Canvas Pane, keeping the Book Pane visible so you can read and write side by side. Click the button again to close.
Rich text editing
The Notebook supports full rich-text formatting:
- Bold, italic, underline,
strikethrough, andinline code. - Three levels of headings (H1, H2, H3).
- Bullet lists, numbered lists, and task lists with checkboxes.
- Block quotes, code blocks (with syntax highlighting), and horizontal dividers.
- Inline links, embedded images, and embedded videos (YouTube, Vimeo).
You can use the floating toolbar on text selection or the standard Cmd/Ctrl+B, Cmd/Ctrl+I keyboard shortcuts.
Slash commands
Type / anywhere in the Notebook to bring up a quick-insert menu:
| Command | Inserts |
|---|---|
/h1 | Heading 1 |
/h2 | Heading 2 |
/h3 | Heading 3 |
/todo | Task list item with checkbox |
/bullet | Bullet list item |
/ordered | Numbered list item |
/quote | Block quote |
/code | Code block with language selector |
/divider | Horizontal rule |
/image | Insert image from URL or upload |
/video | Embed a YouTube or Vimeo video |
/table | Insert a resizable table |
Send to Note
Select any text in a report, then click → Send to Note in the floating toolbar. The selected text is appended to the current project’s Notebook with:
- A source reference showing which page and section it came from.
- Automatic blockquote formatting so it reads as a quoted passage.
- A clickable backlink so you can jump back to the original location.
Autosave
The Notebook saves automatically as you type — there is no manual save button. A small indicator in the bottom-right corner confirms the latest save. If you lose connection, your local edits are preserved and sync as soon as you reconnect.
Export & Settings
Export your work, configure appearance, manage your account and plan.
Export formats
Click Export as in the toolbar to download your report:
| Format | Details |
|---|---|
| Professional layout with serif typography, page numbers, and a table of contents. Paper sizes: A4, A3, Letter, Legal. | |
| Markdown | Plain text with formatting. Citations preserved as reference links. Ready for GitHub, Notion, Obsidian, or static-site generators. |
| DOCX | Microsoft Word format with styles, tables, footnotes, and references. Compatible with Google Docs and Pages. |
The canvas can be exported separately as a PNG image (see Chapter 4).
Theme
Open Settings → Appearance and pick one:
- Light — warm, paper-like background for daytime reading.
- Dark — low-glare dark mode for late-night sessions.
- System — auto-matches your operating system preference.
You can also toggle theme from the sun/moon icon in the top-right corner of any page.
Language
Innogath’s interface is available in nine languages. Open Settings → Language to switch. Supported:
The language setting affects UI labels and menus. Research outputs are generated in whichever language you write your prompt in.
Default chat mode
Choose which chat mode is selected by default when you open a new project:
- Auto — the safe default; routes automatically.
- Fast — best if you mostly do quick brainstorming.
- Thinking — best for analysts who prefer visible reasoning.
- Deep Research — best if nearly every session is a full report.
Font size
Three reading sizes for the Book Pane: Small, Medium (default), and Large. Independent of your browser zoom, so you can keep the sidebar compact while enlarging only the report text.
Account management
Open Settings → Account to manage your profile:
- Change email — sends a confirmation link to the new address.
- Change password — requires your current password for security.
- Enable two-factor auth — TOTP (Google Authenticator, 1Password, Authy).
- Download my data — export all your projects and notebooks as a ZIP.
- Delete account — permanently removes your account and data. This action is irreversible; confirmation is required.
Plans & billing
Innogath offers three tiers:
| Plan | Who it’s for | Credits |
|---|---|---|
| Free | Try the product, run a handful of reports per month | Limited monthly credits |
| Pro | Power users, students, solo researchers | Higher monthly credits + priority processing |
| Ultra | Analysts, teams, heavy Deep Research usage | Maximum credits + all premium features |
See the full comparison on the Pricing page. Manage your subscription in Settings → Billing. Payment processing is handled by our authorised Merchant of Record — see the Refund Policy for details on upgrades, downgrades, and refunds.
Keyboard shortcuts
EnterShift EnterCmd/Ctrl KCmd/Ctrl BCmd/Ctrl MCmd/Ctrl NCmd/Ctrl ZCmd/Ctrl Shift Z/EscapeSupport
- Email: support@innogath.com
- Privacy Policy: innogath.com/privacy
- Terms of Service: innogath.com/terms
- Changelog: innogath.com/changelog
When contacting support, include your account email, project name, and a short description of the issue. Screenshots and a copy of the Activity Stream help us resolve problems fastest.
Innogath