This bounded sample contains twenty-seven fully reviewed posts from twenty-seven external creator accounts. The repeatable promise is a bridge from code or plain-language prompts to an editable artifact: creators generate UML, flowcharts, class diagrams, sequence diagrams, concept maps, and planning diagrams through Mermaid AI, Mermaid Chart's official ChatGPT integration, its web editor, or its GitHub Copilot extension, then refine, export, share, or place the result in documentation.
The strongest example is ForrestKnight's explicitly sponsored developer tutorial, which connects AI generation and code editing to a real GitHub README deliverable. Two more creators disclose affiliate discounts; the other twenty-four relationships remain unknown because product attribution and ordinary links do not prove a commercial tie.
Two external TikTok workflows extend the sample beyond developer tutorials: a Spanish creator turns a real invoice-approval brief into Mermaid code and renders every decision and loop, while an Indonesian thesis educator repairs a research flow with Mermaid Chart AI and exports the final diagram as PNG.
The latest engineering pass adds two product-deep workflows: one turns a software-architecture brief into a generated diagram, simplifies it with Mermaid AI, changes its direction and exports it; the other moves between Mermaid Chart's visual and code views, invokes AI simplification, and repairs a large system graph.
Earlier additions cover the official GPT, Claude-generated architectures, business-analysis diagrams, database ER modeling, iterative prompt repair, Mermaid AI timing, and Gemini CLI output. A custom Workato visualizer, an interview with no product screen, and an unbranded 16-second renderer demo were rejected; tool-list, Edraw.AI, generic syntax, competing renderer, interview-only, and name-only candidates also stayed out.
The original 25-post target is complete and the external TikTok extension has begun without admitting official feeds or generic Mermaid.js tutorials. Platform captions were attempted first; when a usable public-caption transcript was unavailable, bounded local base ASR supplied the transcript layer and every product decision was cross-checked against the full frame set and public metadata.
This is a thin, 100% non-official reviewed sample, not a census or evidence of market share.
PAID3.5×▶101.9K@StudentPrompt
3creators ran it · 3 posts3 of 3 beat 2× their own median · best 101.9K · 1 past 100K
