Clone a famous app screen — pixel-close
Musicians learn by covering songs; designers learn by cloning screens. Rebuild one screen of a well-known app you use in Figma or Canva, then set a multimodal AI hunting the differences you missed — the gaps teach you the craft.
Posted by The AI Internships
The work brief
- 01Screenshot one screen of an app you use daily. Study it: spacing, font sizes, corner radii, icon weights.
- 02Rebuild it from scratch in Figma (free) or Canva — no tracing over the screenshot.
- 03Compare side-by-side and iterate until the differences are hard to spot.
- 04Hand it to a critic: give a multimodal AI both images and ask it to list every difference — verify each claim yourself, fix the real ones, and note where the AI was wrong.
- 05Write the gap notes: measurements you got wrong at first, details you’d never noticed (spacing rhythms, colour subtleties).
What you’ll produce
Submission standard
Submit the side-by-side comparison, the shareable design file link, and your gap notes, plus your AI workflow. Also submit a screenshot of your editor's layers panel with your account and date, and the public share link to your AI difference-hunt chat.
Original screenshot and your clone, side by side
Original and clone side by side — every difference you claim to have fixed in your gap notes must be visibly reconciled here, so the two artifacts tell one consistent story.
Image uploadRequiredPublic share link to your Figma/Canva file
Public share link that opens to a real layered/component build (not an imported flat image). A dead, private, or image-only link fails.
Public linkRequiredWhat you measured wrong at first, and 5 details you’d never noticed before cloning
Written responseRequired
You’ll complete these inside your private workspace.
What you must submit as proof
This brief requires evidence an AI can’t fabricate.
Screenshot of your OWN Figma/Canva editor with the layers/objects panel open — account name and today's date visible
Show the real layer stack of your build with your username and system date — proves construction, not tracing over the screenshot.
Image uploadRequiredPublic share link to your AI difference-hunt chat
The ChatGPT/Gemini 'Share' link where you gave the AI both images and hunted differences — must be YOUR live chat, showing real back-and-forth.
Public linkRequired
Submissions without this evidence cannot be submitted.
Protect other people in your proof. Blur faces, names, phone numbers and email addresses before you upload, and refer to anyone you worked with by role or number ("Listener 1", "the stall owner"). Your proof is only ever used to check your work — it is never published, never appears on your certificate, and is never shown in your public portfolio.
How your work is evaluated
The passing benchmark is 70/100.
Clone accuracy
29%Spacing, type scale, and proportions are convincingly close.
Genuinely rebuilt
14%The file shows real construction (layers, components), not a traced image.
Observational learning
29%The gap notes show genuinely sharpened observation.
Constructed and cross-checked
29%Auto-fail if the file link is dead/private or an imported flat image, if the layers screenshot is missing, or if the differences named in the gap notes are not visibly reconciled in the side-by-side comparison.
How we grade your AI usage
Using AI is the point — it’s the skill this certificate proves. You’ll answer three short questions about how you used it: what you asked, what was wrong with its first answer, and what you changed. Specific, honest answers score high. “I pasted the brief and submitted the answer” scores near zero.