Prompt Cartography
Frame
Clarify audience, purpose, data readiness, and cartographic strategy before asking a model to draw anything.
Start with intentWorkflow
A better map starts before the prompt.
Prompt cartography is the practice of using natural-language prompts as cartographic design briefs.
The cartographer still decides what the map is for, what the data can support, how the map should look and behave, and how its limits should be disclosed. AI helps generate, critique, and revise the work—but it does not replace cartographic judgment.
Casual request
Make a flood risk map.
Prompt cartography adds
Audience, data limits, visual hierarchy, uncertainty, critique, revision.
Defensible map
A map someone can use, question, revise, and cite.
Quick start
The whole method in three moves.
The book expands this spine into a larger design practice: frame the work, make a testable prototype, and prove it is ready to use, question, revise, and hand off.
Make
Turn the design problem into a prompt brief, generate a draft, critique the result, and write targeted revision prompts.
Open the making loopProve
Check accessibility, ethics, uncertainty, publication context, and reproducibility before calling the map finished.
Finish responsiblyFull pipeline
Click a stage. See what it does.
Each stage makes one kind of cartographic judgment visible, reusable, and easier to critique.
Frame
Chapter 1
Intent & audience
Define who the map serves and what decision, memory, or conversation it should support.
What you do
Interview the map idea: audience, setting, decision, emotional tone, device, and what the user should remember.
Why it matters
LLMs default to generic map-shaped output unless you provide purpose, audience, and constraints.
Try this prompt
Act as a cartographic project interviewer. Ask me five questions that clarify the map audience, task, data, risk, and final deliverable.
Skeptic check
If you skip this, the model may optimize for a plausible-looking map instead of a useful one.
Before / after
What changes when the workflow shows up?
The point is not a longer prompt. The point is a map brief with enough judgment in it to be tested.
Make a pretty population map.
The model has to invent the audience, purpose, classification, palette, legend behavior, source note, and standard of success.
Create a county population-density map for regional planners comparing settlement patterns.
Use a sequential color-safe palette, explain class breaks, prioritize regional comparison, cite the data source, and include one uncertainty note about county-level aggregation.
Choose your path
Where should you go next?
Different visitors need different proof. Start with the door that matches your curiosity.
I'm newLearn by doing
Start with beginner labs that turn casual map ideas into critique-ready prompt briefs.
I'm skepticalSee map examplesVisit the Web Mapper GPT Suite homepage for a direct encounter with prompt-driven web map generation.
I'm teaching thisUse the classroom kitFind chapter alignment, rubrics, assignments, discussion prompts, and adoption support.
I'm ready to use itBrowse reusable promptsCopy prompts, schemas, and critique agents organized by workflow stage.
The full method
Want the whole Prompt Cartography stack?
This page is the book's workflow spine. The book expands it across prompt roles and cartographic memory, data pipelines, visual language, color, typography, symbology, thematic mapping, animation, interaction, ethics, and maintenance.