— Door 04 · Capacity, honestly shown

One specialist. A studio's output.

The honest pitch: AI handles the routine half — resizing into every format, generating variants, tagging metadata, renaming and filing — so the human half (brand taste, guideline adherence, catching the placeholder text, final quality) gets all of me. The panel below is an illustrative concept, not live data.

Read the application →
▣ AI does the routine half
  • Resize one asset into every required formatscaled
  • Generate first-pass variants to choose fromdrafted
  • Tag metadata against the annotation schematagged
  • Apply naming convention & file the assetfiled
  • Flag low-confidence / odd text for reviewflagged
♥ The human keeps what matters
  • Does it actually look on-brand?human
  • Catch placeholder / gibberish texthuman
  • Follow the guidelines exactlyhuman
  • Taste, hierarchy & final polishhuman
  • Sign off before it joins the sethuman

Workstream

End-to-end

create → document → file

Real designs built

3

live in Door 02

Quality bar

4 gates

all pass/fail

Placeholder text shipped

0

the only acceptable number

Asset workflow feed · illustrative

Text flag

"The poster came back with AI placeholder text ('Blixel Squad System'). I opened the editing session, read every layer and replaced it with clean copy before it shipped."

Brand check

"Variant 3 drifted off the palette — held it back and corrected the gradient to the brand violet/cyan before filing."

Ready to file

"Three assets resized, tagged with metadata and named to convention — ready for your sign-off and the training set."

How the engagement runs

Phase 1Align on brand, formats & the AI-training label taxonomy
Phase 2Create in Canva · AI resizes & drafts variants, I direct & choose
Phase 3Document · metadata tagged, decisions logged for annotation
Phase 4Review · run the quality bar, catch text/brand issues
Phase 5File & version · name to convention, archive old, hand off
See the studio → See the concept →

A line I won't blur

AI accelerates resizing, variant generation, tagging and filing. It never decides what's on-brand, it never ships unchecked text, and it never replaces taste or guideline judgement. Every asset that joins the set is reviewed and signed off by a human — me. This dashboard is a concept to show the workflow, not a live system.