Skip to main content
Bower can draw two different things in a conversation: figures, which are generated image files suitable for taking away, and mermaid diagrams, which render from a fenced block and suit anything structural. They are different tools for different jobs, and Bower picks based on what you ask for.

​
Figures

A figure is a plot of data: a chart of your results, a comparison across conditions, a distribution. Ask in plain language, and say what you want plotted:
  • “Plot the pH readings from this note against time”
  • “Compare the yields across the three conditions in this table”
Figures come back as real image files rather than as something drawn on screen and lost, so you can take one away and put it where it needs to go.
Point Bower at the note or attachment holding the numbers before you ask, using a context chip or by attaching the file. A figure is only as good as the data it was handed.

​
Mermaid diagrams

A mermaid diagram is structural rather than numerical: a workflow, a decision tree, a sample-handling chain, an experimental design.
  • “Draw the workflow for this protocol”
  • “Diagram how these samples were split across the two arms”
These render inline in the conversation from a fenced block.

​
Which one you get

Ask for a plot of numbers and you get a figure. Ask for a workflow, a process or a relationship between things and you get a diagram. If you get the wrong one, say which you wanted; Bower takes the correction.

​
Limits worth knowing

Bower draws what your data says. It will not invent points to fill a gap, and a figure built from a sparse or ambiguous source is a sparse figure rather than a confident-looking wrong one. Scientific illustration, meaning drawings of apparatus, organisms or anatomy, is deliberately out of scope. Bower refuses those rather than producing something that looks authoritative and is not. A figure is an assistive output. Check it against your source before it goes into anything that matters.