The visual is not decoration: the gap in a diagram, the colour split on a map or the row in a matrix holds the concept being tested — remove it and the question cannot be answered. It does not have to be extracted from your file either; it can be constructed from plain text. A consistency check compares the figure against the question text and flags mismatches for review.
No card required
The figure is generated for the question, not picked from a stock library — so it contains exactly the data the item asks about.
Bar charts, line graphs, pie charts and tables that carry the data the question actually asks about — the core of data-interpretation items.
Triangles, circles, coordinate planes and composite shapes with the labels and measurements the question depends on.
Venn diagrams, process flows, physics setups and labelled scientific diagrams for questions that need a structure, not a photo.
The figure is produced together with the question rather than picked from a bank, so it can be built for the exact distinction being tested — including from a source that contains no images at all.
Upload one PDF, DOCX, PPTX, TXT, JPG, or PNG file. JPG and PNG files are interpreted with AI vision. A scanned PDF with no text layer is detected automatically and processed from its page images.
Choose visual question types and the subject area. Difficulty and cognitive level are set the same way as for text questions.
A visual agent draws the figure while the question is written, then a consistency check confirms the text and the drawing agree — that the labels, values and shapes the question references are the ones actually drawn.
Preview each figure alongside its question and answer key, edit what you want, then export.
It is a tool that produces exam questions together with the figure each question needs — a chart, diagram or geometry drawing — instead of only writing text and leaving you to source or draw the image yourself.
No. If your document contains figures they can be used, but the point is that a figure can be constructed even when the source is plain text — a set of rules can become a rules panel plus a case matrix, for example. That is what makes the visual part of the question rather than an illustration of it.
A consistency check runs after generation and compares the question text against the figure. Pairs that fall below the threshold are flagged for human review rather than passed through silently. It is a review aid, not a guarantee: you confirm the figure before using the question.
Not only mathematics and science. It also works in literature, history, geography and law, where a thematic map, a process diagram, a scene or a case matrix can carry the concept being tested. Data-interpretation items in social science rely on the same mechanism.
You can review every figure with its question before export and regenerate any pair you are not happy with. Questions and answer keys remain fully editable.
Yes. Visual questions use the same configuration as text questions, so you can target a Bloom's Taxonomy level or a PISA or TIMSS framework alignment and still get an item that comes with its own figure.
The first 5 questions are free with no card required, then pay-as-you-go per question with no monthly subscription.