Design living room art layouts
I asked Codex for a living-room art playground, then had to point out overlapping content and poor layout. I’m keeping this conversation as a Personal AI fail: measured drawings and a long list of deliverables did not add up to a good visual design.
Highlights
- I’m presenting this as an AI fail in visual design. The number of deliverables and Codex’s reported cleanup do not make it a successful design.
- Codex’s first comparison contained three schemes, twelve measured wall studies, and four generated room mockups across the requested walls, excluding the bookcase wall.
- After I supplied eleven references, Codex produced two revised schemes. Its descriptions of broader groupings, varied frames, and deliberate whitespace are descriptions of the attempts, not my approval of them.
- The final source checkpoint reports three Figma pages and sixteen artworks with editable image, mat, and frame layers. I’m retaining that record without treating it as proof that the design succeeded.
- I called out the overlapping content directly: “ok, everything is like on top of each other. the layout is not good.” Codex reported reorganizing the file afterward.
Source conversation
This transcript is not public. You can request access below; requests are reviewed before anything is shared.
Created artifacts
My Personal AI fail case study, with measured evidence of the attempted art layouts.
Actions and follow-ups
Actions Taken
Codex read my room measurements, references, and sixteen-artwork library. It produced three initial schemes, twelve measured wall studies, an interactive comparison, a printable guide, and illustrative mockups. After eleven additional references, it produced two revised schemes and a frame-and-mat schedule. A second-pass TV mockup failed; the measured TV elevations remained available.
I then had to flag an obvious problem in the Figma playground: “ok, everything is like on top of each other. the layout is not good.” Codex reported reorganizing it into separate measured, reference, and photo pages and checking dimensions, spacing, and editable layers. I’m recording what it reported, not claiming that this made the visual design good.
Actions Outstanding
The furniture dimensions remain approximate, and some proposed enlargements need higher-resolution source artwork. This record does not establish a final accepted scheme, frame purchases, or installation. This writeup reflects the captured conversation; it does not include a fresh check of the current Figma file.
Source extracts
basically i want a playground
ok, everything is like on top of each other. the layout is not good.
Decisions
- I’m classifying this as Personal and AI fail, and keeping the attempted layouts as evidence.
- I wanted measured elevations to anchor the dimensions, with approximate furniture and generated mockups kept distinct.
- I wanted multiple schemes to compare against my references, with artwork, mats, and frames I could edit independently.
- I’m keeping the original room photographs and full conversation private; the public images are measured elevations.