推介体检

读到第二句时

写给谁看:技术采购方 · 可信度:通篇都是没有支撑的断言

投资人或客户会不会接着读下去?

Jev 读到的 state

Jev is a System One model: it doesn't write prose, it returns typed decisions with calibrated probabilities. Paste any text below and watch nine of them come back in a single request — charted, shareable, and impossible to parse wrong.

9
问题
1
次请求
324毫秒
往返耗时
753
输入 token
US$0.000032
费用
jev-1.13.0
模型

领衔的几条解读

价值是否清楚

Score

一个陌生人读一遍能不能明白。

52%
3.43 满分 4 · 读到第二句时

写给谁看

Choice

这段文字假定的读者是谁。

86%
技术采购方 占 88%,共 5 个选项
88% 技术采购方

可信度

Score

这些主张有没有被撑住。

87%
0.16 满分 4 · 通篇都是没有支撑的断言

同一次请求里的其余部分

差异化

Score

竞争对手能不能直接复制粘贴这段话。

86%
2.92 满分 4 · 一个清晰、具体的差异

用了真数字

Noul

增长、定价、规模。

几乎肯定是
答案为「是」的概率
95%
抛硬币

点明了问题

Noul

开药方之前有没有先说痛点。

确实说不准
接近抛硬币——别只凭这一个分支
46%
抛硬币

行话过重

Noul

用品类黑话顶替实质内容。

偏向是
答案为「是」的概率
67%
抛硬币

有明确诉求

Noul

读者接下来该做什么。

确实说不准
接近抛硬币——别只凭这一个分支
49%
抛硬币

会收到回复

Noul

冷邮件的试金石。

偏向否
答案为「是」的概率
35%
抛硬币

分享这条解读

拿到链接的人都能看,并且会出现在「浏览」里。

发到 X
产生这条结果的确切请求

POST https://api.typesafe.ai/v1/systemone

{
  "state": "Jev is a System One model: it doesn't write prose, it returns typed decisions with calibrated probabilities. Paste any text below and watch nine of them come back in a single request — charted, share…",
  "model": "jev-latest",
  "questions": {
    "value_clarity": {
      "type": "score",
      "instructions": "How quickly does a reader understand what this offers?",
      "criteria": [
        "Never — the value is never stated",
        "Only after several reads",
        "By the end of the paragraph",
        "By the second sentence",
        "Immediately, in the first line"
      ]
    },
    "audience": {
      "type": "choice",
      "instructions": "Who is this pitch written for?",
      "criteria": {
        "Investors": "Framed around market, traction, and returns",
        "Technical buyers": "Framed around how it works and integrates",
        "Business buyers": "Framed around outcomes, cost, and risk",
        "Consumers": "Framed around personal benefit",
        "Nobody in particular": "Too generic to identify a reader"
      }
    },
    "credibility": {
      "type": "score",
      "instructions": "How well are the claims in this pitch supported?",
      "criteria": [
        "Unsupported assertions throughout",
        "Mostly claims, one soft proof point",
        "A mix of claims and evidence",
        "Most claims carry specifics",
        "Every claim has a number or a name behind it"
      ]
    },
    "differentiation": {
      "type": "score",
      "instructions": "How clearly does this distinguish itself from alternatives?",
      "criteria": [
        "Any competitor could have written it",
        "A vague gesture at being different",
        "A stated difference, unproven",
        "A clear, specific difference",
        "A difference that is hard to copy and easy to verify"
      ]
    },
    "has_numbers": {
      "type": "noul",
      "instructions": "Does the pitch contain specific numbers?"
    },
    "names_the_problem": {
      "type": "noul",
      "instructions": "Does the pitch clearly state the problem it solves?"
    },
    "jargon_heavy": {
      "type": "noul",
      "instructions": "Is the pitch obscured by jargon?"
    },
    "has_ask": {
      "type": "noul",
      "instructions": "Does the pitch make a clear ask of the reader?"
    },
    "would_reply": {
      "type": "noul",
      "instructions": "If this arrived cold, would the intended reader reply?"
    }
  }
}

把 Jev 指向你自己的文本

九个类型化答案,一次请求,大约半秒。挑一个镜头,或者自己写问题。