Tutorials · Chapter B (2/4) · ~9 min
Multimodal prompts
Try it → see it → read → next
Text, images, and voice are different inputs into the same helpful job.
Try yourself
Playground
Mix the senses
Turn modalities on/off. Multimodal = more than one channel into the same ask.
Enable image or audio with text, then run.
Recap
What you just did
You practiced describing an image clearly — subject, style, don’ts. The same clarity helps when you upload a picture and ask for analysis.
Teach
How it works
Copy-paste patterns:
Photo → text
Role: careful tutor
Task: Explain what this diagram is doing in plain English
Context: I’m a beginner; avoid jargon
Format: 5 bullets + one “try this” step
[attach image]
Screenshot → fix
Task: Find the error in this UI screenshot
Context: Users tap Save and nothing happens
Format: Likely causes (3) + what to check first
Voice notes work the same — speak Role + Task + Format, then attach the clip if the app allows.
Use it
When you'd use this
- Homework photo → step hints (not full answers if school forbids).
- Recipe photo → shopping list.
- Whiteboard photo → typed action items.
Watch out
Watch out
Don’t upload IDs, medical charts, or private work docs. Vision models can misread text in images — verify important numbers.
Try next
Try this next
Photograph a short handwritten list and ask an AI to type it as bullets. Check accuracy.