Prompt Playground

This page runs your prompt against a live model. Pick a goal, edit the prompt, then hit Run prompt and the page sends it to the model and shows the answer below. One goal asks the model to draw a chart, which the page renders beside a known-good reference chart so you can compare. There is also a floating tutor pill for follow-up questions.

The Prompt Library gives you patterns for driving analysis with AI. This is where you practice them. Below is a small customer table. Pick an analysis goal, watch the matching prompt pattern fill in, edit it to make it your own, and run it against this data.

The data: 20 customers

A synthetic accounts-receivable extract. Revenue is right-skewed, two accounts are far larger than the rest, and there are two segments. This is the table every prompt below runs against.

customer_idsegmentrevenue ($)days_to_payregion
Shaded rows are the two outliers, the accounts far larger than the rest.

1. Pick an analysis goal

Each goal is one of the five moves from the Prompt Library. Click one and its prompt pattern fills the box below, already adapted to this table's columns.

2. Edit the prompt

Why this prompt works. Pick a goal above to see the pattern and why it earns a good answer.
The good edits: make it more specific, name the exact column, state what the data looks like, and ask the AI to show its reasoning. The note above tells you what this pattern is going for.

3. Run it

Run prompt sends your prompt straight to the model, which gets the 20 rows along with it, and shows the answer below. If you would rather use your own AI in another tab, Copy prompt gives you the text plus the data to paste in. Either way, compare what you get back to the model answer.

Copied
Model answer
Show what a strong answer looks like

Pick a goal first. This reveal shows the kind of answer a strong version of that prompt should pull back, so you can compare it to what you got. It is a yardstick, not a substitute for running your own.

The moves, at a glance

The first five cards are the five categories from the Prompt Library page, each pre-filled with a worked version pointed at this table. The sixth, Chart it, asks the model to draw a chart of the same data so you can compare it to the reference chart the page already knows is right.

A prompt earns a good answer when it is specific, hands the AI the shape of the data, asks for reasoning rather than just a number, and, for a claim, makes the AI argue against you. Never paste a number, or a chart, back into a slide without checking it yourself.