PDF ⛶
🌻 Map Panel#
🗺️ What you can do here: See your causal relationships as an interactive network map. Drag nodes around, click on links to edit them, and use the controls to customize how the map looks. You can even drag one factor onto another to quickly create new links. This is where your data comes to life visually.

Map Controls #

Map Legend #

Discrete text legend showing:

Evidence (shortcut to Links Print View)#

Map Formatting #

Customisable formatting (Things you can tweak)#

Layout and interaction

Factors

Links

Other

Fixed visual appearance (things you can't tweak)#

Some parts of the map’s appearance are automatic (i.e. they are not controlled by the Map Formatting widgets above):

Leading emoji (Print / Graphviz only):

Link geometry (bundling):

Automatic colouring overlays:

Automatic highlighting:

Interactive Features#

These work for all layouts except Print/Graphviz layout (which is mostly for static export, but does support clicking nodes/links now).

Editing and deleting (multiple) factors#

What does "everywhere or in current view only" mean?

💡Tip: By control-clicking or shift-clicking multiple factors you can easily rename several at once, e.g. you can merge multiple factors as a single factor.

Grid layout#

What this is for: Sometimes you want factors to appear in a rough grid—like “this cause is clearly to the left of that one”—instead of leaving the layout algorithm to guess. You do that by putting a small coordinate tag in the factor label (anywhere in the text). The app reads the tag to place the box, and (by default) hides the tag so respondents don’t see a string of numbers.

This is especially useful when:

How to write a tag: Use two whole numbers: a column along the main flow of the map and a row across the flow. For example (2,1) or (2.1) in the label means “second step along the story, first slot across.” You can use square brackets instead of parentheses, and you can leave one side blank if you only care about one direction—for example (3,) fixes the step along the flow but not the slot across.

In Map Formatting

Interactive map (the usual live map): tagged factors snap to a grid; anything without a tag is still laid out automatically, but is kept near the grid so the picture stays readable.

Print / Graphviz (static SVG): the app can’t pin exact pixel positions like the live map, but it still respects left‑to‑right order of the first number along the direction you chose, and does its best to respect the second number within each step. If you only give a second number and not the first, Print layout can’t place that factor on the “step” axis—use both numbers when you care about order.

Heads-up: The grid helpers only apply to the default interactive layout and to Print. If you switch the live map to layouts like Cola, Circle, or CiSE, coordinate tags are not used for placement.

Vignettes #

📝 What you can do here: Generate AI-powered narrative summaries of your causal maps. Choose between a **whole-map** summary that covers the relationships in your current filtered view, or a **typical-source** story that focuses on one representative respondent. Useful for reports, annexes, and explaining your analysis in plain language.

For practitioners (getting started)#

What are vignettes and why use them?
Vignettes turn your coded causal map into readable prose. After you have links (and usually quotes on those links), the AI can draft a narrative that names themes, tensions, and patterns in your material. You stay in control: you choose the prompt (tone, audience, structure), and you can edit the text afterwards. They are a fast way to move from “many links on a map” to “something I can paste into a report or share with stakeholders”—not a substitute for your judgement, but a structured first draft.

Whole-map vignette
Use this when you want a bird’s-eye story of everything that is currently in the map (respecting your filters and source selection). The app sends a compact summary of factors and bundled links (with sentiment), evidence snippets (quotes and source IDs from highlighted bundles where available), and up to 30 “typical” sources scored by how much they represent common bundles—each with bundles, link-level quotes, and a small metadata preview (title, filename, and simple custom fields). If you want the narrative to emphasise particular edges, set Map Formatting → Links highlight to Significant or Feedback loop first; that snapshot is included so the model can focus there. AI region (where Gemini runs) follows Project Details → AI region, not a separate control on this card.

Typical-source vignette
Use this when you want a single-respondent case study: the app picks the most “representative” source for the current map view (using link counts and coverage of bundles, with a bias so very long documents do not always win). It sends that source’s full text plus its links with quotes and sentiment, so the model can write as if telling one story. Do not ask it to generalise across the whole project in this mode—the prompt and data are scoped to that one source.

How to use (UI):

  1. Open the Vignettes card on the Map side panel and expand it if needed.
  2. Choose model (and optional thinking controls if your model supports them). Region for AI is set per project under Edit project (AI region), not on this card.
  3. (Optional) Leave Enable checking (second AI pass) on so a checker can review the draft; notes appear in a collapsed panel when relevant.
  4. Edit the Whole-map prompt and/or Typical-source prompt (prompt history: dropdown and prev/next; see tips on using prompt history).
  5. Click Write Whole-map Vignette or Write Typical-source Vignette (two separate buttons).

Tip (optional): tell the AI which links matter most (whole-map mode):

What each mode sends (summary):

Output format: Markdown (headers, bold, lists, blockquotes, code blocks). You can copy from the result area.

Bookmarking & restore: