Prompt your presentation
Run your app successfully in the editor, then select Prompt your presentation next to Quick suggestion. Describe the output you want in up to 300 characters. Choose Cheap and fast or Slow but better. The displayed price is half the corresponding app creation price, rounded up to a whole credit. One prompt pays for all suggestions, including cancelled or failed generations. Sign in and check your balance before entering a prompt.
The timeline shows progress and the activity log follows a suggestion that is still working. Your administrator can enable one to five parallel suggestions. Only suggestions that pass the calculation and presentation checks appear for review.
Select Suggestion 1, Suggestion 2, or another available result to view its output at 65% scale and inspect its presentation code. Accept this suggestion applies that presentation and runs the app. Close leaves your code unchanged. The calculation model stays the same. Undo and redo restore both code files and run the restored version.
Ask explicitly for a language when needed. Otherwise headings and help follow the app's language, falling back to your prompt's language.
Use DataCards for independent summary cards from published records, RecordChart for age or other numeric record rows, and UncertaintyChart for actual simulation percentiles alongside the deterministic base case. Their complete reference and examples describe supported fields, units and selections.
For income, expenses and their difference, ask for P50 amounts in every card, chart, table and breakdown. These widgets can share a sampled calculation source that selects the same income and expense metrics. Derive the positive gap as max(0, P50 expenses - P50 income); this is a difference of medians. Ask for only the two income and expense lines with no uncertainty band. Before simulation, the widgets display the published deterministic values with an explanatory note. Missing simulation values stay unavailable.
Explanatory labels
Use Label to add short formatted help alongside results:
help: Label({
title: false,
text: "**Reading the forecast**\n\nThe chart shows projected values based on your current inputs.",
}),
summary: MetricCards({ title: "Your forecast" }),Labels support headings, bold, emphasis, inline code, paragraphs and bullet lists. HTML and links are displayed as text.
Text that follows input changes
The text property also accepts a JavaScript function returning Markdown. It runs after each calculation with the current input values. For example, if your root input is domain and it contains family.members:
funding: Label({
title: "Proportional funding",
text: ctx => "**How it works**\n\n" +
ctx.input.domain.family.members
.filter(member => member.role === "adult")
.map(member => member.name)
.join(" and ") +
" contribute to the household in proportion to their salaries.",
}),Changing a person's name and pressing Play updates the explanation. Each scenario gets its own text, shown when that scenario is selected. Markdown formatting is supported in both string and function forms.
ctx.input uses the same root names as your declared preconditions. A root declared as preconditions: input.domain(...) is available as ctx.input.domain; preconditions: { workspace: input.domain(...) } uses ctx.input.workspace. Legacy inputs such as preconditions: { salary } use ctx.input.salary. Sections, groups and objects keep their nesting; object arrays contain ordinary row objects. Money values have amount and currency, for example ctx.input.salary.amount.toLocaleString("sv-SE").
You can also read published calculation metrics, for example ctx.metrics["metric.savings"], and the scenario name in ctx.scenario. Read existing values rather than repeating financial calculations in the explanation. Monte Carlo runs include the ordinary base calculation as well as simulation summaries. KPI cards, statements and text use the base calculation; distribution widgets show the sampled outcomes. Explain that these are different views of the same assumptions.
Use a synchronous function returning a string, with at most 8,000 characters. Arrow functions and function(ctx) { return ...; } are supported. Input and metric values are read-only. Titles remain static strings. Use ordinary string formatting, conditional expressions and array methods such as filter, map and join; avoid side effects or asynchronous work.
Editable values and fixed assumptions
Never write editable input values into static help, including people's names, dates, amounts and percentages. Read them through a text function when relevant, or keep the explanation generic. This applies to every input type, even if an initial value looks realistic.
Facts fixed in the calculation code outside input can be explained as fixed assumptions. This helps people understand limits that changing input will not affect. Keep explanations in everyday language and focused on the purpose of the app.