Demande une relecture
Le lecteur irait au bout : Probablement pas · Densité de jargon : Dense en mots à la mode
Neuf lectures d’un brouillon, avant de l’envoyer.
Le lecteur irait au bout : Probablement pas · Densité de jargon : Dense en mots à la mode
Neuf lectures d’un brouillon, avant de l’envoyer.
发生了什么: @aleyda 转发 @CyrusShepard 对 130+ 从业者的调研(9/18):品牌信任与 parametric memory 权重大于 chunking / llms.txt;可引用的原始数据胜泛文;传统可抓取/可排名与 AI 可见高度重叠。@lilyraynyc 在深圳场强调:改叙事要打全表面,Reddit 单人长帖就能毁品牌在模型里的画像;有 Knowledge Panel 可即时验证 Google 资料。GSC 同时被「伪对话式 prompt 追踪词」刷屏。
Si un lecteur qui découvre le texte sait ce qui est dit.
Mots à la mode par paragraphe.
La seule mesure d’engagement qui compte.
Affirmations précises contre abstractions.
Le niveau d’études que la prose présuppose.
La voix que le lecteur entend.
Ce qui fonctionne déjà.
Des phrases supprimables sans rien perdre.
Si quoi que ce soit de réfutable est avancé.
Toute personne disposant du lien peut la voir, et elle apparaît dans Explorer.
POST https://api.typesafe.ai/v1/systemone
{
"state": "发生了什么: @aleyda 转发 @CyrusShepard 对 130+ 从业者的调研(9/18):品牌信任与 parametric memory 权重大于 chunking / llms.txt;可引用的原始数据胜泛文;传统可抓取/可排名与 AI 可见高度重叠。@lilyraynyc 在深圳场强调:改叙事要打全表面,Reddit 单人长帖就能毁品牌在模型里的画像;有 Knowledge P…",
"model": "jev-latest",
"questions": {
"clarity": {
"type": "score",
"instructions": "How clear is this writing to a first-time reader?",
"criteria": [
"Impenetrable",
"Needs re-reading",
"Followable with effort",
"Clear",
"Effortless"
]
},
"concreteness": {
"type": "score",
"instructions": "How concrete and specific is this writing?",
"criteria": [
"Pure abstraction, no specifics at all",
"Mostly general claims",
"A mix of specifics and generalities",
"Mostly specific",
"Every claim is grounded in a specific detail"
]
},
"jargon": {
"type": "score",
"instructions": "How heavy is the jargon and buzzword usage?",
"criteria": [
"Plain language throughout",
"Occasional industry terms",
"Noticeably jargon-heavy",
"Dense with buzzwords",
"Almost entirely buzzwords"
]
},
"reading_level": {
"type": "score",
"instructions": "What reading level does this writing assume?",
"criteria": [
"Primary school",
"Middle school",
"High school",
"Undergraduate",
"Specialist or academic"
]
},
"register": {
"type": "choice",
"instructions": "What voice is this written in?",
"criteria": {
"Conversational": "Sounds like a person talking to you",
"Professional": "Polished and neutral",
"Corporate": "Institutional, hedged, committee-written",
"Academic": "Formal, citation-shaped, careful",
"Promotional": "Selling, superlative-heavy"
}
},
"strongest_element": {
"type": "choice",
"instructions": "What is the strongest thing about this draft?",
"criteria": {
"The opening": "It earns the next sentence",
"The structure": "It is easy to navigate and follow",
"The evidence": "Claims are supported by specifics",
"The voice": "It sounds like a person with a point of view",
"The ending": "It lands and tells you what to do",
"Nothing": "No element stands out yet"
}
},
"would_finish": {
"type": "noul",
"instructions": "Would an average reader finish this?",
"criteria": {
"true": "They would read to the end",
"false": "They would stop partway"
}
},
"has_filler": {
"type": "noul",
"instructions": "Does this contain sentences that could be deleted without losing meaning?"
},
"says_something": {
"type": "noul",
"instructions": "Does this make a specific claim a reader could disagree with?",
"criteria": {
"true": "It takes a position or states a checkable fact",
"false": "It is unobjectionable and could have been written about anything"
}
}
}
}Neuf réponses typées, une requête, environ une demi-seconde. Choisissez une lentille ou écrivez vos propres questions.