BrandVoice校對
A B2B copy-polishing service for Taiwan marketing teams that uses the same de-AI rule set to enforce a company's approved vocabulary, terminology, and tone across every AI-assisted draft.
Marketing leads at Taiwan SaaS and e-commerce companies whose teams draft with AI but ship inconsistent copy
- Company glossary and forbidden-term list imported from a style guide
- Batch review of marketing emails, landing pages, and ad copy with diff view
- Team dashboard showing which writers lean on which AI patterns most
- Slack and LINE integration for in-flow approvals before publish
Internal AI policy is now a board-level concern; teams need a controlled way to keep AI drafts inside house style without hiring extra editors.
Open-source 繁中去 AI 味 skill is on GitHub Trending (original MrGeDiao/shuorenhua) and K-media explicitly frames 內容精修 as a brand-voice/professional-image concern for AI-generated content, confirming real pain — but B2B-paid-service demand in Taiwan is not yet evidenced.GitHub - tentenco/shuorenhua-zh-tw (Taiwan fork of trending skill) ↗Ai內容精修術：守護品牌語調與提升專業形象 - K-media ↗开源史上最强去AI味Skill - Pluvio9yte on X ↗
Most 去 AI 味 skills (humanizer, shuorenhua, speak-human) target Simplified Chinese; the Taiwan-specific fork currently has 0 stars/forks on GitHub, and no competitor offering per-company approved-vocabulary enforcement for Traditional Chinese surfaced in searches.GitHub - tentenco/shuorenhua-zh-tw ↗去AI味十大Agent skill - 博客园 (PetterLiu) ↗speak-human-tw - SkillsLLM ↗
Core de-AI capability is already MIT-licensed open source, compressing the wedge a paid wrapper can charge for; no public pricing for Taiwan-specific B2B copy-polishing surfaced, and the K-media piece is thought-leadership rather than evidence of budgeted spend — willingness-to-pay is unverified.GitHub - tentenco/shuorenhua-zh-tw (MIT license) ↗Ai內容精修術 - K-media (content piece, no pricing data) ↗
Internal AI governance is becoming a board-level concern in Taiwan (2026 AI Basic Act framework, data-sovereignty discourse), and brand-voice consistency will persist as long as teams draft with AI — but underlying de-AI logic risks commoditization as base models improve at matching house style.台灣 AI SaaS 生態系的合規挑戰 - Webslots blog ↗GitHub - tentenco/shuorenhua-zh-tw (explains why de-AI matters long-term) ↗
The core skill (scene detection, protected spans, Taiwan-localization rules, 60-pt scoring) already exists in open source and can be wrapped with a per-tenant terminology/tone rule layer; main build cost is B2B GTM and onboarding, not the underlying tech.GitHub - tentenco/shuorenhua-zh-tw (existing skill architecture) ↗