Skip to content

Templates and A.I. prompts

A spectator gives you one piece of information — a word, a card, a name, a date. Templates and A.I. prompts turn it into something else: a whole sentence, a zodiac sign, a famous quote, a drawing. The result goes to your display devices, the printer, or the next step of your routine.

All three work the same way: you write texts in advance (with variables from the Text Buffer), each gets a number (#1, #2...), and an action runs the one you need.

Feature What it makes Action
TEMPLATES a text, from your own sentence text from template
A.I. TEXT PROMPTS a text, written by the A.I. execute A.I. text prompt
A.I. IMAGE PROMPTS a picture, drawn by the A.I. execute A.I. image prompt

They are in Settings → Transform / process, and need the Transformer add-on. The A.I. prompts also need the A.I. Features add-on and an A.I. engine (your own API key, or the free Gemini calls of the subscribers). All add-ons are included in every subscription.

The variables

Templates and prompts use the same variables:

Variable Meaning
#TEXT the last text in the Text Buffer
#TXT1 .. #TXT9 counted from the oldest: #TXT1 is the first text
#LAST1 .. #LAST9 counted from the newest: #LAST2 is the one before the last

Add !uc (UPPERCASE), !lc (lowercase) or !tc (Title Case): #TEXT!uc, #TXT2!tc.

Only when everything is there

A template or prompt runs only if the Text Buffer has enough texts for its variables: one with #TXT3 waits for the third text — until then, the action does nothing. So you can put it on Action after text recognition, and it will work only after the third word. More: Text Buffer variables.

Managing the lists

Each feature has its own list (Manage templates → Open list, Text prompts → Open list, Image prompts → Open list):

  • Add a new one by typing it into the last field.
  • Edit one by simply overwriting its text.
  • Swipe right on one to delete it.
  • The number in front of each (#1, #2...) is what the actions use. Deleting one doesn't change the numbers of the others.
  • Save and close.

The lists are saved with your routine.

Templates

Your own sentence, with the spectator's information inserted.

Example: "In my dream I saw the #TXT1 in the hands of #TXT2."

A spectator picks any card and places it under the card box... another spectator writes the name of their favorite celebrity on a small notepad. Then you show them your handwritten prediction on a chalkboard: "In my dream I saw the 7♠ in the hands of Natalie Portman."

  • The action text from template (Special group) asks for the Template #ID.
  • The result goes into the Text Buffer (as a template result), so it is also the new #TEXT — send it to a device with show last text, print it, or use it in a prompt.
  • Action after template generated + Trigger action: what happens next (e.g. handwriting to printer), with a delay you choose. The phone vibrates (long) when it runs.

A.I. text prompts

The same idea, but the A.I. writes the answer.

Example: "What is the zodiac sign of #TEXT? Please only reply the name of the zodiac, with all uppercase letters."

A spectator writes the name of their favorite celebrity on a small notepad (for example Brad Pitt). Then you show them your handwritten prediction on small, crumbled paper: "SAGITTARIUS"

More ideas: "Name a famous painting with the color #TEXT", "Write a two-line poem about #TXT1 and #TXT2", "In which year was #TEXT born? Reply only the year."

Settings in A.I. TEXT PROMPTS:

  • A.I. text engine: OpenAI GPT, Google Gemini or Claude. The model itself is chosen in A.I. FEATURES.
  • Show response (pop-up): shows the answer (and how long it took) on your screen.
  • Action after response received + Trigger action.

The action is execute A.I. text prompt (Special group, asks for the Prompt #ID). The answer goes into the Text Buffer (A.I. prompt response).

Short answers

MysterSmith asks the A.I. to answer as short as possible, only the result, in the language of your prompt. Still, say exactly what you want: "Reply only one word", "only the number", "all uppercase letters".

A.I. image prompts

Describe a picture, and the A.I. draws it.

Example: "A simple black and white child's drawing of #TEXT, on white paper, no text."

Settings in A.I. IMAGE PROMPTS:

  • A.I. image engine: OpenAI GPT or Google Gemini (their drawing settings — size, model, thicker lines — are in A.I. FEATURES).
  • Action after image generated + Trigger action — runs when the picture is ready and already in the Image Buffer, so immediately is fine.

The action is execute A.I. image prompt (Special group, asks for the Image prompt #ID). While the A.I. is drawing, a connected PeekSmith shows Drawing.... The picture appears on the black screen and the Doodle screen, and is saved to the Image Buffer — from there you can send it to a printer or a display, or put it into an image template.

From other apps

The HTTP Socket and the custom URL scheme can run them too: template=N, aiprompt=N, aiimageprompt=N (N is the number in the list).

See also: Acronym word list — another Transformer feature that makes a word list from the spectator's word.