BeforeOriginal AI-generated demo
Meaning Translator
Turn vague feedback into clear design direction.
Choose the reaction or build concern that matches what you are seeing. Add your context and constraints, then copy a detailed prompt into your AI design or coding tool.
20 design reactions · 9 implementation concerns · editable promptsWhat does the design feel like?
These are intentional starting points, not an automated diagnosis. Select the closest vibe and the complete prompt builder will open with your choice clearly identified.
Message + priority
Layout + experience
Back to the topBrand + audience
Back to the topTrust + quality
Back to the topTone + energy
Back to the topDeveloper implementation
Back to the topThese are build problems, not visual vibes. Use them when the concept is directionally right but the design is incomplete, ambiguous, fragile, inaccessible, or difficult to translate into production code.
Tested on a working design
See the Meaning Translator in action
I tested the Translator on an existing AI-generated law-firm demo using Meta AI. I supplied the live working page, completed the Translator fields, and pasted the generated prompt without additional design coaching. The first pass did not simply restyle the page. It translated credibility into a more specific editorial structure, real legal context, clearer practice boundaries, and audience-aware attorney profiles.
AfterMeta AI first pass using the prompt
This is the raw first pass, not a finished design. It is evidence that one structured prompt moved a working page away from a conventional generated template and toward a distinctive, context-specific foundation. Open the After page and scroll to the attorney section to toggle each profile between Corporate and Family Law views.
Behind the library
How the Meaning Translator was built
The Meaning Translator was built from a much deeper body of work than a typical prompt library. Its foundation includes the original 43-page Semantics of Design course, the Signal, Meaning, Action framework, public research collected from 2025 through today across Reddit, GitHub, Hacker News, Substack, design communities, developer forums, and product discussions, plus more than two years of my own prompts, feedback, corrections, notes, and implementation instructions from real design and development projects.
This material was distilled into structured feedback records connecting the original reaction to its likely meaning, underlying design principle, recommended action, preservation constraints, common overcorrections, and verification criteria. Similar reactions were consolidated, designer-centric terminology was rewritten in language non-designers naturally use, and newer AI-specific problems were added, including generic output, lost context, unnecessary redesigns, missing interface states, design-system drift, responsive failures, and implementations that no longer match the intended experience.
The result is a customizable prompt library covering both stakeholder feedback and developer implementation problems. Users select a recognizable concern, add their project context and visual requirements, then generate an editable prompt that tells an AI what to inspect, what to change, what to preserve, what mistakes to avoid, and how to verify the result. The current library is the first usable layer of a larger context engine grounded in this expanding research corpus and years of accumulated design judgment.
Semantics of Design + SMA
Public research, 2025 to today
2+ years of project feedback
Structured meaning layer
Custom prompts
Bounded, testable action
Bounded, testable action