人们实际上如何使用 Jev

Jev是我们很长时间以来最受关注的模型之一，因为它不仅仅是像GPT-6、Opus或Fable那样的又一个LLM。

Jev is one of the buzziest models we've had in a long time, and that's because it's not just another LLM like a GPT-6 or an Opus or Fable model.

它是根本不同的东西。

It is something fundamentally different.

但正因为它的不同，一开始并不完全清楚它最适合用来做什么。

But because it's different, it's not necessarily clear at the beginning exactly what it's going to be best used for.

不过，经过一个半星期的实践，人们正在发现并分享各种不同的用例，这些用例利用了Jev的独特之处，今天我们就来聊聊其中最好的那些，以及它们可能对你有什么帮助。

With the benefit of a week and a half under our belts now, though, people are discovering and sharing a slew of different use cases that take advantage of what makes Jev unique, and today we're going to get into the best of them and where they might be relevant for you.

《AI每日简报》是一档每日播客和视频节目，内容涵盖人工智能领域最重要的新闻和讨论。

The AI Daily Brief is the daily podcast and video about the most important news and discussions in AI.

好了，朋友们，在我们开始之前，先快速说几个通知。

All right, friends, quick announcements before we dive in.

首先，感谢今天的赞助商：KPMG、Blitzy、Harbor和Hyperagent。

First of all, thank you to today's sponsors, KPMG, Blitzy, Harbor, and Hyperagent.

想要收听无广告版本，请访问patreon.com/aidailybrief，或者你也可以在Apple Podcasts上订阅。

To get an ad-free version of the show, go to patreon.com/aidailybrief, or you can subscribe on Apple Podcasts.

如果想了解如何赞助这个节目，请给我们发邮件到sponsors@aidailybrief.ai。

And to learn more about sponsoring the show, send us a note at sponsors@aidailybrief.ai.

还有一个快速通知。

One more quick announcement.

我们即将举办下一场免费网络研讨会。

We have our next free webinar coming up shortly.

这次研讨会的内容是如何构建你自己的个人AI基准测试，这样当新模型发布时，你可以测试它，看看它对你来说有多好，以及它在你的人工智能工具链中适合什么位置。

It's all about how you can build your own personal AI benchmark so that when a new model comes out, you can test it and see how good it is for you and where it will fit into your AI stack.

这次研讨会还是由Nufar Gaspar主持。

It's led once again by Nufar Gaspar.

研讨会是免费的，你可以在aidailybrief.ai获取所有需要的信息。

It will be free, and you can get all the info you need at aidailybrief.ai.

过去几周，人工智能领域最热门的新事物之一是一个名为Jev的新模型。

Over the last couple of weeks, one of the buzziest new things to come up in the AI world has been a new model called Jev.

Jev的有趣之处在于，它不仅仅是另一个用于与GPT-6或Opus 5.5相同用途的大语言模型，而是实际上以略有不同且互补的方式工作，我们稍后会看到。

Now, what makes Jev interesting is that it is not just another LLM that you would use for the same thing as GPT-6 or Opus 5.5 but actually works in a slightly differently, and as we will see, complementary way.

自发布以来的大约10天里，不仅有很多热议，我说的是X上数百条不同的帖子，每条帖子本身又有数百甚至数千的点赞和分享，而且这种关注也转化成了巨大的财务机会，有信息报道说该公司正在洽谈以100亿美元或更高的估值筹集高达10亿美元的资金。

In the 10 days or so since launch, not only has there been a ton of buzz, I'm talking hundreds of different posts on X, which each themselves have hundreds or even thousands of likes and shares, but that attention has also translated into significant financial opportunity with the information reporting that the company is in talks to raise as much as $1 billion at a $10 billion or higher valuation.

这相比它完成的4000万美元种子轮融资（当时估值为2亿美元）来说，是一个相当大的飞跃。

That is a decent jump from its $40 million seed that was completed at a $200 million valuation.

但现在既然大家有机会真正上手使用Jev了，我想回过头来聊聊，除了社交媒体上那些花哨的视觉和游戏演示之外，人们实际上是怎么用这个东西的。

But now that we've had a chance for people to actually get their hands on Jev itself, I wanted to go back through and talk about how people are actually using this thing outside of the buzzy visual and game demos that have been all over social media.

基本上，Jev是不是一个不是游戏设计师或开发者的普通人应该关注，甚至考虑纳入自己更大AI工具组合的东西？

Basically, is Jev something that the average person who's not a game designer or not a developer should be paying attention to and even thinking about as part of their larger AI stack?

现在，来回顾一下Jev是什么，我之前称它为判断模型。

Now, to recap what Jev is, previously, I called it a judgment model.

Jev不能做的是以你通常认为聊天机器人写作的方式来写作。

What Jev can't do is write in the traditional way that you think about chatbots writing.

相反，构建Jev的Typesafe团队称它为系统一模型，这是基于丹尼尔·卡尼曼《思考，快与慢》中的一个概念。

Instead, the team at Typesafe who built Jev calls it a system one model after a concept from Daniel Kahneman's "Thinking, Fast and Slow."

系统一思维是快速的、直觉性的模式匹配，也就是瞬间判断，而Jev正是为此而设计的。

System one thinking is fast, instinctive pattern matching, i.e. snap judgments, and that's what Jev is built for.

一个判断某项任务是否适合Jev的好方法是，看它是否涉及反复阅读内容、做出小判断，然后采取可预测的下一步行动。

A good way to think about a test for whether something is a good job for Jev is where you repeatedly read something, make a small judgment, then take a predictable next step.

比如，一个文件出现在你的下载文件夹里。

So take, for example, a file landing in your downloads folder.

现有软件可能会用一条普通规则，把它归类为PDF。

An ordinary rule that existing software might have would be to classify it as a PDF.

但Jev能更进一步，识别出它是不是发票、属于哪个项目、是否需要有人查看。

But what Jev can do is go a step further, identifying is it an invoice, which projects is it for, and does someone need to see it?

一旦做出判断，就可以按系统流程进入下一步。

Once the judgment is made, it can be predictably moved on to the next step in a system.

你可以向Jev提出三类核心问题。

There are three core types of questions that you can ask Jev.

第一类是“选一个”，也就是Jev所说的“选择”。

The first is pick one or what Jev calls choice.

它会回答哪个选项更匹配。

It answers which of these fits.

你可以列出最多255个选项，Jev会帮你找到最合适的。

You list up to 255 options, and Jev can find the best fit.

举个例子，比如这个工单应该由哪个团队处理？

So an example of this might be which team should handle this ticket?

是账单、技术支持、销售，还是其他？

Is it billing, tech support, sales, or other?

Jev模型会返回你选择的那个选项，同时还会给出每个其他选项的概率。

The Jev model is going to give you back the pick, plus a probability for every other option.

第二种可以问Jev的问题类型是他们所谓的评分，基本上就是按一个等级来打分。

The second type of question that you can ask Jev is what they call a score, which is basically rating on a scale.

你用文字描述每个等级，从2级到10级，Jev就能判断某个东西属于哪个等级。

You describe each level in words from two to 10 levels, and Jev can answer where something falls.

比如，从平静到非常生气这个等级上，这个客户有多不满？

So for example, on the scale from calm to very angry, how frustrated is this customer?

你会从Jev那里得到一个在这个等级上的位置，以及模型对这个判断有多确定的指示。

You're going to get back from Jev a position on that scale as well as an indication of how sure the model is.

最后一种问题是是非题，Jev 称之为 null，这是 Bernoulli 的简称，它只是回答：这是真的吗？

The last type of question is a yes or no question, what Jev calls a null, which is short for Bernoulli, and it simply answers, is this true?

继续用这个客服的例子，这个人是不是明确要求退款？

So to stay on this customer service example, is this person explicitly asking for a refund?

你会得到一个从 0 到 1 的概率，表示这句话为真的可能性。

What you're going to get back is a probability from zero to one that the statement is true.

X 上的 HeyStefan 做了一个非常酷又简单的可视化，他在屏幕底部放了一堆表情符号，然后在文本框里输入了“你可以输入分类器，比如可以穿的东西，或者冬天可以穿的东西。”

HeyStefan on X did a really cool, very simple visualization where he put a pile of emojis at the bottom of a screen, and then into a text box, he wrote, "You can type in classifiers like things you can wear or things you can wear in winter."

匹配的表情符号会跑到顶部，如果分类变了，它们就会掉下来。

The emojis that match race to the top or fall off if the classification has changed.

在后台发生的是，每个表情符号都在回答同一个是非题。

And so what's happening in the background is that every emoji is getting the same yes or no question.

这符合你输入的内容吗？

Does this match what you typed?

重要的是，没有人像传统机器学习那样给表情符号打标签或做标记。

And importantly, nobody labeled or tagged the emoji as would happen in traditional machine learning.

换句话说，根本没有关于冬天的规则。

There's no rule, in other words, for winter.

每个表情符号在同一时刻都会收到同样的问题：这个匹配吗？

Every emoji gets the same question at the same moment, does this match?

如果你在看这个演示，你可能会注意到，因为Jev做的事情有限且具体，所以它运行起来非常快。

And part of what you might be recognizing if you're watching this demo is that because what Jev does is limited and specific, it can do it incredibly fast.

Typesafe发布这个模型时，他们说它的运行速度比类似的大语言模型流程快20到200倍，成本低40到400倍。

When Typesafe launched the model, they said it can work 20 to 200 times faster and 40 to 400 times cheaper than comparable processes with LLMs.

现在，不要把这堆表情符号想象成表情符号，而是把它想象成你的潜在客户、你的工单、你的文档，或者其他一大堆没有区分的信息。

Now, instead of this pile of emojis, imagine that the pile is your leads, your tickets, your documents, or some other big undifferentiated mass of information.

重要的是，你可以同时问同一个项目很多问题。

Importantly, you can ask many questions about the same item all at once.

因为问题是并行运行的，问第九个或第十个问题不会花更多时间，只是多消耗一些token。

Because questions run in parallel, asking the ninth or 10th question doesn't cost more time, it just costs more tokens.

比如说，你有一封客户邮件。

Say, for example, you have a single customer email.

你可能想同时问三个是或否的问题。

You might want to ask three yes or no questions at once.

客户生气了吗？

Is the customer angry?

他们是否得到了延迟回复？

Did they get a delayed response?

他们是不是在处理错误的产品？

Are they dealing with an incorrect item?

这三个是独立的问题，每个都会有各自的概率。

These are three separate questions that are each going to have their own probability.

在他们的测试中，Typesafe发现，一次调用处理13个问题比逐个处理便宜12.2倍，快10倍，而且得到的答案完全相同。

In their testing, Typesafe found that 13 questions handled in a single call was 12.2 times cheaper and 10 times faster than one at a time and got to the identical answers.

Matthew Berman分析了724个实时广告，每个广告被问了12个问题。

Matthew Berman did an analysis of 724 live ads, with each ad being asked 12 questions.

一个关于钩子原型，一个关于格式，一个关于报价，一个关于行动号召的意图，一个关于认知度，等等。

One about the hook archetype, one about the format, one about the offer, one about the CTA intent, one about the awareness, et cetera.

他发现，针对那条广告的12个问题，单次请求只用了173毫秒就完成了。

He found that that single request for the 12 questions for that ad took just 173 milliseconds to complete.

而这正是模型强大之处所在。

And this is really where the power of the model comes in.

它不仅仅擅长判断，而且速度快、成本低，可以对所有事情进行提问和判断。

It's not just that it's good at judgment, it's that it's fast enough and cheap enough to ask and make judgments about everything.

一百万输入标记只需4.2美分，输出标记免费。

A million input tokens cost just 4.2 cents, and output tokens are free.

每次调用大约需要70到500毫秒，而且正如我们所见，可以在一次调用中组合多个不同的问题。

And each call is going to take between 70 and 500 milliseconds, and as we've seen, can combine a bunch of different questions in a single call.

重要的是，Jev在性能上做了取舍，这使它擅长某些任务，但也意味着它不会做某些事情。

Now, importantly, Jev makes trade-offs that allow it to be good at the things that it does, but that also means that it doesn't do certain things.

它不会写代码，不会起草合同，也不会做出涉及多种不可量化、不明确因素的复杂决策。

It's not going to write code, it's not going to draft contracts, and it's not going to make nuanced decisions that involve a variety of factors that aren't quantifiable and clear.

它将被用于大规模的小型判断任务。

It's going to be used for small judgments at volume.

哪个类别？

Which bucket?

有多紧急？

How urgent?

相关吗？

Is it relevant?

安全吗？

Is it safe?

现在我们来聊聊人们实际使用Jev的六种方式，而不只是那些酷炫的演示。

So now let's talk about six different ways that people are actually putting Jev to work outside of just cool demos.

第一类是分析你已有的内容。

A first category is analyzing what you already have.

第二类是按语义搜索。

A second category is searching by meaning.

第三类是对进来的内容进行分类处理。

A third category is triaging what comes in.

第四类是检查你的工作是否符合你的规则。

A fourth category is checking work against your rules.

第五类是加速AI代理，第六类是即时响应。

A fifth category is speeding up AI agents, and a sixth category is responding instantly.

那我们来看第一类，分析你已有的东西，也就是“这堆里有什么？”

So let's move to category one, analyzing what you already have, AKA what's in this pile?

基本上就是拿你已有的一个集合，对每个项目问Jev同样的问题，然后统计答案。

This is basically where you're going to take a collection you already have, ask Jev the same questions about every item, and then count the answers.

这种分析以前靠手工做是不划算的，但现在只需要几秒钟，成本也就几分钱。

This is analysis that previously could not be justified doing by hand but is now going to take seconds and cost cents.

一个例子就是我们刚提到的Matthew Berman，他分析了37个品牌的724条在播广告，拆解出每个钩子、每种格式、每个优惠、每个行动号召。

One example was that from Matthew Berman that we just talked about, where he analyzed and broke down 724 live ads from 37 brands to identify every hook, every format, every offer, every CTA.

这样做的目的是拿到这些数据后，跟实际表现好的广告做对比，从而获得非常深入、细致、精细的广告表现信息，这在以前是极其困难的。

The idea is to be able to then take this data and compare it to what actually performed, giving you incredibly deep and complex fine-grain information about advertising performance that would've been extremely difficult before.

Berman拿到那724条在播广告的完整拆解，只花了40秒，token成本9美分。

It took 40 seconds and 9 cents worth of tokens for Berman to get the full breakdown of those 724 live ads.

两天后，伯曼更进一步，这次他没有泛泛地问那些问题，而是让Jev以30种不同的买家身份浏览了723条广告。

Two days later, Berman went further, and instead of just asking those questions in general, this time he had Jev scroll 723 ads as 30 different buyer personalities.

这些身份是像健身房老板、牙科诊所经理、幼儿妈妈或对AI好奇的工程师这样的典型形象。

The personalities were archetypes like gym owner, dental office manager, toddler mom, or AI curious engineer.

Jev的任务是，对每个买家和每条广告，判断这个人会停下来看还是继续往下滑。

Jev's job was to ask for every buyer and every ad, would this person stop scrolling or keep going?

花了22美分，就得到了21690个“停下还是滑走”的判断。

It cost 22 cents to get 21,690 stop or scroll decisions.

当然，这不算真正的数据。

Now, of course, this is not data.

这不是30个真实的买家。

This wasn't 30 actual buyers.

这只是AI想象中的买家典型。

It was just buyer archetypes as considered by an AI.

换句话说，最好把它看作一个假设生成器，但很难不想到，营销会转向把它作为流程中的关键部分，在花钱做真实测试之前，先用来检验广告或落地页的角度。

In other words, it's better to view this as a hypothesis generator, but it's hard not to think that marketing will shift to using this as a key part of its process to pressure test ad or landing page angles before paying for real tests.

很多人都在用Jev处理整个档案库，不管是邮件存档、过去的X帖子、文档文件夹，还是研究项目的数据库。

Lots of other folks are running Jev over whole archives, whether that's email archives, past X posts, folders of documents, or databases of research projects.

在每种情况下，Jev做的事情都是问同样几个问题，但针对档案中的每一个条目。

And in each case, what Jev is doing is asking the same few questions, but about every single item in the archive.

比如Ian Nuttall，他让Jev对将近3300条他们之前的X帖子，每条问八个问题，包括主题、吸引点和语气，然后把这些数据与互动指标对比，更好地了解什么真正有效。

Ian Nuttall, for example, had Jev ask eight questions each, topic, hook, and tone, about nearly 3,300 of their previous X posts to be able to then compare that data with engagement metrics to get a better sense of what actually worked.

顺便说一下，这3300条帖子，每条八个问题，总共花了大约13美分。

And by the way, those eight questions each across 3,300 posts cost about 13 cents.

那我们怎么尝试类似的事情呢？

So how might we try something like this?

基本上，你手头任何一大堆数据，比如一年的客户邮件、CRM笔记、一个会议记录文件夹，任何你能写出几个简单问题并针对所有数据运行的地方，都是Jev任务可能派上用场的地方。

Basically, any big pile of data you have, a year of customer emails, CRM notes, a folder of meeting transcripts, anything where you can write a few simple questions and run them against all of those queries, that's a potentially useful place for a Jev job.

第二类是语义搜索。

Category two is searching by meaning.

换句话说，哪些内容符合我的意思？

In other words, which of these match what I mean?

这个思路就是，你用大白话描述你想找的东西，然后让Jev去检查每一个候选。

The idea is to be able to describe what you're looking for in plain words and let Jev check every candidate.

它的厉害之处在于，即使关键词对不上，它也能找到你真正想要的东西。

The power is that it can find what you mean even when the words don't match.

这方面有很多例子，基本上就是人们把它当作一种新的网站自然语言搜索方式。

There were a bunch of examples of this, with people basically using it as a new approach to natural language search on websites.

a16z的Justine Moore举了个例子，扫描了数千条Zillow房源，然后按一些通常没法筛选的条件来分类，比如建筑风格、翻新状态，或者离高速路的远近。

Justine Moore from a16z gave the example of scanning thousands of Zillow listings and classifying properties by things you can't normally filter for, such as architectural style, renovation status, or proximity to freeways.

有个例子特别吸引我，Burhan拿了一个90分钟的视频，然后根据他用自然语言描述的主题来剪辑片段。

In one example that piqued my interest, Burhan took a 90-minute video and was able to clip it based on themes that they described in natural language.

比如，在那个90分钟的视频里，他搜索并剪辑了“他们对AI何时会自动化AI研究的预测”，结果不到两秒就找到了好几个相关片段，成本还不到两美分。

So for example, across that 90-minute video, they were able to look for and clip, quote, "Their predictions for when AI will automate AI research," and they were able to find a bunch of examples of that in under two seconds for under 2 cents.

另一个例子是SEO审计，Borja让Jev读完了他们网站的全部586个页面，重新构建了网站的内部链接地图，整个过程只用了45.1秒，花了21美分。

Another example of this is an SEO audit, where Borja deployed Jev to read all 586 pages of their website to rebuild the site's internal link map, which it was able to do in 45.1 seconds for 21 cents.

而Claude Opus 5只读了其中21个页面，就花了1.43美元。

Claude Opus 5 only got through 21 of those pages and spent a buck 43.

Borja解释说，内部链接正是Jev最擅长的工作。

Borja explains, internal linking is the perfect Jev job.

这不是写作。

It's not writing.

这是8790个是或否的判断。

It is 8,790 yes or no calls.

这个页面有没有真正的理由链接到那个页面？

Does this page have a real reason to link to that one?

而且文案里是不是已经有锚文本了？

And is there anchor text already sitting in the copy?

这是一个分类问题，而我们一直在为逐页处理支付高昂的费用。

That is a classification problem, and we have been paying frontier prices to do it one page at a time.

按意义搜索的最后一个子类别是根据你关心的内容来筛选你阅读的东西。

A final subcategory of the searching by meaning is filtering what you read by what you care about.

比如，Robin Bilgic构建了一个实时AI垃圾内容检测器。

So for example, Robin Bilgic built a real-time AI slop detector.

当他们在X上滚动时，这个垃圾内容检测器会给出一个从0到1的置信度分数，表示它认为某条帖子是AI垃圾内容或不是，并屏蔽掉达到一定置信度阈值的帖子。

As they scroll X, the slop detector gives a confidence score on that zero to one scale about how much it thinks a post is AI slop or not, blocking out the ones that reach a certain confidence threshold.

不过，这种过滤方式可以应用于很多不同的场景。

This sort of filtering, though, can be applied in a bunch of different ways.

Elvis San在X上举了一个例子，让Jev浏览384条早间新闻，并识别出其中15个不同品牌应该关注的新闻。

Elvis San on X gave an example of letting Jev browse 384 morning news stories and identifying which of them 15 different brands should be paying attention to.

这个任务在24.9秒内完成，花费19美分，而相比之下，在同一时间段内，Opus 5只读完了其中4篇文章，剩下380篇未读。

That was completed in 24.9 seconds for 19 cents, and by way of comparison, in that same time period, Opus 5 got through four of those articles, leaving 380 unread.

Jev的第三个用例类别是对收到的内容进行分类。

Jev use case category three is triaging what comes in.

换句话说，就是判断这是什么，以及它应该被放到哪里。

In other words, what is this and where does it go?

人们正在尝试的问题包括：这封邮件有多重要？

The kinds of questions that people are experimenting with are things like, how important is this email?

这个下载的文件是发票吗？

Is this downloaded file an invoice?

这个链接是不是恶意的？

Is this link malicious?

这个线索应该给销售、自助服务还是培育？

Should this lead go to sales, self-serve, or nurture?

有一个特别有趣的实验来自乔纳森·尤尼科夫斯基。

One really interesting experiment came from Jonathan Yunikowski.

他写道：“每个邮件应用都按时间倒序显示收件箱。

He wrote, "Every email app shows your inbox in reverse chronological order.

如果改成按重要性实时排序呢？”

What if it was live prioritized by importance instead?"

在这个案例中，Jev的任务是给每封邮件的重要性打分，测试中它能在453毫秒内给100封邮件打分，成本大约十分之一美分，乔纳森说Jev的打分和他自己的判断完全一致。

Jev's job in this case was to rate every email's importance, and in the test, it was able to rate 100 emails in 453 milliseconds for about a 10th of a cent, with Jonathan reporting that Jev's rating matched his own on every single one.

这个功能我现在就想要，不是明天。

This is a use case that I want right now, not tomorrow.

其实我昨天就想要了，那些因为没收到我回复而抓狂的人也都想要。

In fact, I want it yesterday, and so do all of the people who are sitting there beating their heads against the wall because I haven't responded to them yet.

还有一些人在尝试一种模式，就是每个项目问一个快速的问题。

Some other folks are experimenting with a pattern of asking one quick question per item.

Beyond Code的首席技术官Marcel Pokio写道：“我构建了一个macOS应用，用来监控我的下载文件夹，并配有一套可自定义的规则。

Marcel Pokio, the CTO of Beyond Code, writes, "I built a macOS app that monitors my downloads folder along with a customizable set of rules.

下载的文件是发票吗？

Is the downloaded file an invoice?

如果是，就把它移到专门的文件夹，并改成正确的文件名。”

Move it to a special folder with the correct file name."

没有调用其他大语言模型，只用Jev。

No other LLM calls involved, just Jev.

DevEd用Jev来做实时聊天审核，一旦出现脏话或负面评论就立即删除。

DevEd used Jev for live chat moderation, removing swearing and negative comments as they arrive.

Dub短链接服务的Steven Tay把之前捕获的10,000个恶意域名喂给Jev，用来在他们的免费短链接服务上标记不良链接。

Steven Tay of the Dub link shortener fed Jev 10,000 malicious domains they'd caught before to flag bad links on their free shortener service.

他写道：“这是我们从第一天起就一直在努力解决的问题。

He wrote, "This has been something that we've been wrestling with since day one.

有了Jev，我们两个小时就搞定了。

With Jev, we solved it in two hours."

其实，这种分类处理在很多业务中都非常关键。

In fact, this sort of triage is so integral to so much of business.

在一篇关于营销人员使用Jev的10个用例的文章中，Umon X把工作流决策列为一个用例，他说：“大多数工作流最终都会遇到同一个问题：接下来该做什么？”

In a post about 10 Jev use cases for a marketer, Umon X identified workflow decisions as one, saying, "Most workflows eventually hit the same question, what should happen next?

这可能是使用Jev的最佳场景。

That's probably the best place to use Jev."

Box举了一个例子，他们已经尝试过用Jev进行事件分类，用来判断客户影响和严重程度，以及可以升级到哪些路径。

Box gave an example, having already experimented with incident triage, where Jev is used to judge customer impact and severity, as well as things like which routes it can be escalated to.

Odoo CRM有一个提议的模块，对每个新线索问三个问题：优先级、购买意愿，以及是否是垃圾信息。

Odoo CRM has a proposed module that asks three things about every new lead, their priority, their buying readiness, and whether it's spam or not.

如果你只能选一个领域来探索，尤其是当你的公司里有多个人接触相同的线索、客户或联系人时，我认为这种分类和路由将是Jev变得不可或缺的地方，基本上从它被集成到系统的那一刻起就会如此。

If you had to look at just one area to explore, especially if you are in a company with multiple people touching the same leads or customers or contacts, this sort of triage and routing is, I think, where Jev is going to become absolutely integral, basically from the moment that it gets integrated into the systems.

毕马威和德克萨斯大学奥斯汀分校的一项新研究发现，当人们与AI合作时，相似的技能并不能保证相似的结果。

A new study from KPMG and the University of Texas at Austin found that when people work with AI, similar skills don't guarantee similar outcomes.

研究人员研究了500多名早期职业人士，发现表现最好的人通过引导、评估和完善AI的输出，持续放大了AI的价值。

Researchers studied more than 500 early career professionals and found that the best performers consistently amplified the value of AI by guiding, evaluating, and refining its outputs.

这些顶尖表现者，被称为AI放大器，他们的定义不仅仅在于他们知道什么，更在于他们如何与AI合作。

These top performers, called AI amplifiers, weren't defined by what they knew alone, but by how they worked with AI.

想了解更多关于AI放大器与其他人的区别，请访问kpmg.com/us/aiamplifiers。

Learn more about what separates AI amplifiers from everyone else at kpmg.com/us/aiamplifiers.

Blitzy对庞大代码库的理解，能够实现自主安全修复、现代化改造和新功能开发。

Blitzy's understanding of massive code bases unlocks autonomous security fixes, modernization, and new features.

那么，当完全没有遗留代码时，会发生什么呢？

So what happens when there's no legacy code at all?

绿地项目本该是简单的部分，一切从零开始，没有技术债务。

Greenfield is supposed to be the easy part, clean slate, no technical debt.

但即使是绿地项目，也只能以人类的速度推进，一个冲刺接一个冲刺。

But even Greenfield moves at human speed, one sprint at a time.

Blitzy将工作单元从开发者转变为整个项目，自主规划、构建、测试和验证整个应用，从零开始生成数十万行生产级代码。

Blitzy changes the unit of work from the developer to the project, autonomously planning, building, testing, and validating entire applications from scratch, hundreds of thousands of lines of production-ready code.

Blitzy的一个客户上线了一个全新的应用，53万4千行代码，把原本65周的开发计划压缩到了两周。

One Blitzy customer stood up a brand-new application, 534,000 lines of code, compressing a 65-week roadmap into two weeks.

另一个客户在没有前端工程师的情况下，交付了整个应用。

Another shipped an entire application with no front-end engineer.

无论是老系统还是全新项目，答案都一样：软件以计算的速度运行。

Legacy or greenfield, the answer is the same, software at the speed of compute.

在blitzy.com开始构建你的下一个项目。

Build what's next at blitzy.com.

网址是B-L-I-T-Z-Y.com。

That's B-L-I-T-Z-Y.com.

每期节目，我们都会讨论OpenAI、Anthropic、SpaceX AI、谷歌和Meta之间的竞争。

Every episode, we cover the competition between OpenAI, Anthropic, SpaceX AI, Google, and Meta.

你可能已经对谁领先有了自己的看法，但每个AI实验室都在采取不同的策略，开发不同的技术，建立不同的合作伙伴关系，并打造独特的生态系统。

Chances are you've already formed an opinion about who's leading, but every AI lab is taking a different approach, building different technologies, forging different partnerships, and developing a unique ecosystem.

Harbor Capital Advisors的AI实验室生态系统ETF系列，为投资者提供了一种投资他们认为最有成功潜力的AI生态系统的方式。

Harbor Capital Advisors AI Lab Ecosystem ETF Suite gives investors a way to gain exposure to the AI ecosystem they believe is best positioned for success.

在您投资的任何地方搜索Harbor AI Lab Ecosystem ETFs，或在X上关注@HarborCapital了解更多信息。

Search Harbor AI Lab Ecosystem ETFs wherever you invest or follow @HarborCapital on X to learn more.

请访问harborcapital.com获取包含投资目标、风险、费用、支出及其他重要信息的招股说明书。

Visit harborcapital.com for a prospectus containing investment objectives, risks, fees, expenses, and other important information.

投资前请仔细阅读并考虑这些信息。

Read and consider it carefully before investing.

风险包括本金损失及与人工智能相关的风险。

Risks include principal loss and artificial intelligence-related risks.

Harbor ETFs由Foresight Fund Services, LLC分销。

Harbor ETFs are distributed by Foresight Fund Services, LLC.

Harbor与AI Daily Brief无关，这些基金与任何AI实验室无关联、未获其赞助或认可。

Harbor is not affiliated with AI Daily Brief, and the funds are not affiliated with, sponsored by, or endorsed by any AI lab.

这是一则付费广告，并非个性化投资建议。

This is a paid advertisement and not personalized investment advice.

投资涉及风险，包括可能损失本金。

Investing involves risk, including possible loss of principal.

本期《AI每日简报》由Hyperagent赞助播出，你可以在上面运行多组智能体，让团队一起管理。

This episode of the AI Daily Brief is brought to you by Hyperagent, where you run fleets of agents your team can manage together.

别再依赖本地智能体和聊天工作流，等着你的笔记本电脑来提示了。

Forget local agents and chat workflows waiting on your laptop to be prompted.

Hyperagent在云端部署常驻智能体，在你团队已经使用的工具中完成实际工作。

Hyperagent deploys always-on agents in the cloud, doing real work across the tools your team already uses.

营销智能体把竞争对手的动态变成落地页。

Marketing agents turn competitor moves into landing pages.

销售智能体丰富潜在客户信息、起草邮件，并更新CRM。

Sales agents enrich leads, draft emails, and updates the CRM.

运营智能体处理文书工作，跟踪预算。

Ops agent chases the paperwork and tracks the budget.

每个智能体都能访问共享上下文，并遵守你关于范围和审批的规则。

Every agent has access to shared context and follows your rules about scope and approvals.

是时候让智能体像队友一样加入你的团队了。

It's time you add agents that feel like teammates.

去Hyperagent招你的代理吧。

Hire yours at Hyperagent.

在hyperagent.com/aidailybrief领取100美元积分。

Get $100 in credits at hyperagent.com/aidailybrief.

Jev用例的第四类是检查工作是否符合规则，也就是说，是否达标。

A fourth category of Jev use cases is checking work against rules, i.e., does this meet the bar?

与其用标准的LLM提问方式，比如“请审查这个”，不如把“请审查这个”变成具体的问题，然后对每份草稿都问这些问题。

Instead of the standard LLM question of, "Please review this," you turn, "Please review this" into specific questions and then ask those questions about every draft.

LangChain就有一个例子，他们用同样的方式给AI代理的工作打分。

One example of this came from LangChain, which was grading an AI agent's work the same way every time.

LangChain的Harrison Chase指出，Jev“非常适合做评估，尤其是线上评估，因为你想给大量轨迹打分。”

Harrison Chase from LangChain pointed out that Jev is, quote, "Great for evals, especially online evals, where you want to grade lots of traces."

这起初可能让人觉得更像是给开发者用的，而不是给其他类型的知识工作者和公司用的，但考虑到我们所有人都将开始把代理投入生产，或者管理已经存在的代理，比如支持机器人、研究代理等等，拥有一个更好的工具来构建评估系统，用来评判这些代理的工作，似乎很容易成为核心基础设施。

Now, this might feel initially like something that is more for devs than for other types of knowledge workers and companies, but given how much all of us are going to start putting agents into production or managing agents that already exist, think support bots, research agents, et cetera, having a better tool to build evaluation systems into how we judge those agents' work seems like it could very easily become core infrastructure.

Every团队展示了如何利用这种大规模、超高速检查工作是否符合规则的能力，来改进AI写作。

And the team at Every showed how you could use this sort of ability to check work against rules at mass scale and at an incredible speed as a way to improve AI writing.

他们在12段文字里埋了错误，Jev抓到了七个中的六个，而Claude Fable 5.1抓到了全部七个。

They planted mistakes on 12 passages of writing, with Jev catching six of seven, as compared to Claude Fable 5.1's catching all seven.

不过，Jev在0.35秒内抓到了六个，而Fable 5.1用了8.83秒，而且成本大约便宜了580倍。这意味着你可以反复运行同样的检查很多次，仍然比用前沿模型做同样的审查更便宜、更快。

Jev, however, caught it six in 0.35 seconds as compared to 8.83 seconds for Fable 5.1, and it did so at about 580 times cheaper, meaning that you could rerun that same check a huge number of times and still have it be both cheaper and faster than using a frontier model for the same sort of review.

那么，你实际上怎么把它变成自己能用的东西呢？

So how might you actually turn this into something that you would use?

你基本上需要把你的写作规则转化为一个流程。

You'd basically need to go through a translation process for your rules for writing.

假设你有一份风格指南，或者一份你永远不想看到的短语列表。

So let's imagine that you had a style guide or a list of phrases you never want to see.

你可以把每一条都变成一个“是或否”的问题，然后在你下一篇长文档的每一段上运行这些“是或否”的问题，以确保你的关键沟通中没有AI腔或其他写作雷区。

You could turn each of those into a yes or no question and then run those yes or no questions on every paragraph of your next long document as a way to ensure no AI-isms or other writing third rails in your key communications.

我觉得，这触及了我们需要用Jev进行的最大的思维转变之一。

And this gets, I think, at one of the biggest rewirings that we're going to need to do with Jev.

Jev的独特价值很大程度上不仅在于能做事，而在于能以如此大的规模做事，以至于它实际上变成了质的区别，而不是量的区别。

A lot of the unique value of Jev is not just being able to do a thing, it's being able to do a thing at such scale that it actually becomes a difference in kind rather than a difference in scale.

能够逐句细致地检查每一句话有没有AI腔，这跟对整个文档跑一遍通用的大模型检查，完全是两码事。

Being able to realistically check every single sentence in minute detail for AI-isms, in other words, becomes categorically different from just running a generic LLM check across the document as a whole.

第五类用例，很多人也在尝试，说实话确实有点偏向开发者领域了，但我觉得为了完整性，还是值得聊一聊。

The fifth category of use cases that lots of people were experimenting with admittedly does get a little bit closer to the developer realm, but I think at least for the sake of completeness, it's still worth discussing.

这类用例可以概括为加速你的AI代理，很多内容都围绕我们过去几个月一直在讨论的模型路由。

These use cases you might sum up as speeding up your AI agents, and a lot of this is around the sort of model routing that we've been talking about for the past several months.

大家问的问题大概是这样的：哪个模型能处理这个任务？

The kinds of questions people were asking were things like, which model can handle this task?

这一步需要多少推理能力？

How much reasoning does this step need?

如果有的话，哪个技能适合这个请求？

Which skill fits this request, if any?

这个旧工具的输出还有效吗？

Is this old tool output still relevant?

你可以看到，在这些场景里，共同点就是人们用这些小型自动化的微判断，来引导代理到合适的智能水平、合适的上下文、合适的技能、合适的工具，以最高效的方式完成它的任务。

You can see how in each of these cases, the common thread is people using these small automated micro-judgments to route an agent to the right level of intelligence, the right context, the right skills, the right tools to do whatever its job is in the most efficient way possible.

V Chen写道：“人们用Jev在任务前选择模型。

V Chen wrote, "People use Jev to pick a model before a task.

我让它能在Codex中动态调整GPT-6的推理强度。

I made it change GPT-6 reasoning's effort inside Codex during the task.

遇到困难时多思考，常规步骤则少花力气。”

More thinking when stuck, less for routine steps."

在他们的测试中，发现Astra成本降低了50%，运行速度也更快了。

And in their test, they found 50% lower Astra costs while also getting faster runs.

人们还用Jev来管理上下文窗口。

People are also using Jev to manage the context window.

Daniel Son为Claude Code构建了一个叫Jev Skill Suggestion的工具，他说：“对于每个请求，Jev会判断哪个技能最匹配任务，然后只将该技能注入Claude的上下文中。”

Daniel Son built something called Jev Skill Suggestion for Claude Code, where, quote, "For every request, Jev classifies which skill best matches the task, then injects only that skill into Claude's context."

这使令牌和成本减少了88%。

This led to an 88% decrease in tokens and cost.

我知道，即使你们中很多人以前不是开发者，现在也已经熟练使用Codex和Claude Code，积累了大量技能库。这类基于Jev的工具可能是一种方法，避免每次请求都把所有这些库发送给模型。

I know a lot of even you formerly non-developers have started to be sufficiently proficient with things like Codex and Claude Code that you've built up big skills libraries, and these sorts of Jev-based tools are potentially a way to stop sending all of those libraries to the model with every single request.

这里有一点值得注意，虽然我们现在讨论的很多用例目前还只是个人实验，但你也会看到Jev和类似的判断模型被直接内置到我们使用的工具和框架中。

And one thing to note here is that while a lot of these use cases that we're discussing are, at this stage, individual experiments, you're also going to see Jev and judgment models like it built natively into the tools and harnesses we use.

比如AJ Asver写道：“我们用Jev构建了一个新的框架，把重复性工作的成本降低了90%。

AJ Asver, for example, wrote, "We built a new harness using Jev that cuts the cost of repetitive work by 90%.

这个框架在运行过程中会不断学习，把步骤从大语言模型调用转移到代码中。”

The harness learns the job as it runs, moving steps from LLM calls to code."

他们举的例子是合规警报，他们逐批测量每个警报的成本，每批包含50个合规警报。

The example they gave was compliance alerts, and they measured the cost per alert batch by batch, with each batch representing 50 of those compliance alerts.

一开始，每个警报的成本大约是2.95美元，但到第1000个警报时，成本降到了只有0.25美元。

At the beginning, the cost per alert was about $2.95, but by alert 1,000, it was down to just $0.25.

Jev用例的第六类是关于即时响应的。

A sixth category of Jev use cases are about instant response.

这个人现在想要什么？

What does this person want right now?

很多人都在尝试某种形式的这类应用。

A lot of people were experimenting with some version of this.

马库斯·洛写道：“如果复制粘贴能变得智能会怎样？

Marcus Lowe wrote, "What if copy/paste was smart?

复制一份简历，粘贴到申请表里，字段就自动填好了。”

Copy a resume, paste into an application, and the fields fill themselves."

X上的诺曼也做了类似的事情，把粘贴的文本拆分成片段，识别每个字段，进行匹配，对照原文检查，只粘贴那些高置信度的匹配项。

Norman on X also did a version of this, splitting the pasted text into pieces, working out what each field is, matching them, checking against the original, and pasting only the confident matches.

考虑到工作中很多任务就是把信息从一种格式，比如邮件、PDF、会议记录等，搬到该去的地方，比如CRM、录入表单或模板，这类用例感觉非常实用。

Given how much work at work is moving details from one format, i.e., emails, PDFs, meeting notes, et cetera, into other places where that information is supposed to live, like CRMs, intake forms, or templates, this is a category of use cases that feels very relevant.

现在，知道Jev哪里有用同样重要的是知道哪里需要小心使用。

Now, just as important as knowing where Jev is useful is knowing where you got to be careful with it as well.

我觉得需要格外谨慎的几个领域包括招聘、金钱和安全。

A couple of places that I think warrant greater caution are areas like hiring, money, and security.

在招聘方面，你可以想象，如果被大量申请淹没，你可能会想用Jev来排序，但问题是，即使你给了它好的标准，Jev返回的只是一个数字，没有附带任何推理。

On hiring, you can see how, if deluged by a set of applications, you might be tempted to use something like Jev to rank them, but the problem is that even if you've given it good criteria, Jev is going to return with a number that doesn't have any reasoning attached.

顺便说一句，这也是一个很好的例子，说明你可能需要构建一个更复杂的系统，使用多种类型的人工智能。

This, by the way, is a great example of where you might want to build a more complex system that uses multiple types of AI.

想象一下，你用Jev做同样的排名，但然后自动让其他类型的LLM来审查，比如那些可能值得再看一眼的边缘候选人。

Imagine you do that same sort of ranking with Jev, but then automatically have other types of LLM review, for example, for bubble candidates that might be deserving of a second look.

Jev的开发者Typesafe其实也试图明确说明他们认为Jev不擅长什么。

Typesafe, the makers of Jev, actually even tried to be clear about what they think Jev is bad at.

他们列出的一些问题包括多步骤问题，他们发现每一步准确率都会下降。

Some of the things they put include multi-step questions, where they see accuracy drop with each hop.

他们指出，Jev不太擅长计数、数学或日期，它可以提取事实，但实际的数学计算你最好在其他地方做。

They point out that Jev isn't really good at counting math or dates, that it can extract the facts, but that you're going to want to do the actual math elsewhere.

他们还指出了其他一些问题，比如一致性问题以及理解意图的问题。

And they also point out some other issues like problems with consistency and problems with reading intent.

所以，如果你想判断一个任务是否适合用Jev，四个标准可能会有所帮助。

So if you're trying to figure out if a task is a good fit for Jev, four criteria might help.

第一个是你能提前写下答案。

The first is that you can write down the answers in advance.

想想类别、是或否的二元选择、或者一个评分等级。

Think categories, a yes or no binary, a scale.

如果答案是句子或计算，那可能就不太适合Jev这种判断模型。

If the answer is a sentence or a calculation, that's probably not a good fit for Jev's sort of judgment model.

第二个标准是关于数量的。

Criteria number two is about volume.

就是有一堆或者源源不断的任务，比如几百个工单、几百封邮件之类的。

There's a pile or a stream, such as hundreds of tickets, hundreds of emails, things like that.

第三个标准是关于风险的，就是答错了代价低，或者容易发现错误。

A third criteria is about stakes, where a wrong answer is cheap, or it's easy to catch.

基本上，虽然我把它叫做判断模型，但如果答错后果很严重，你就不想完全依赖它的判断。

Basically, although I'm calling it a judgment model, you don't want to leave things up to its judgment alone if getting it wrong has big consequences.

最后一个标准是，你能不能把需要的证据以文本形式提供给它，而且不超过32,000个token。

A final criteria is whether you can give it the evidence that it needs as text in under 32,000 tokens.

一旦你找到了合适的任务，下一步就是写好问题。

Once you've figured out a good task, the next step is going to be to write good questions.

从这些不同的例子中总结出的一些技巧包括：每个问题只问一个判断。

Some of the tips there from across all of these different examples are things like one judgment per question.

换句话说，一个不太好的问题就是：这是个好线索吗？

In other words, a not so good question is, is this a good lead?

因为那其实不是一个判断，而是把一大堆不同的判断混在一起了。

Because that's actually not one judgment, that's a whole bunch of different judgments embedded in one.

好线索可能指的是行业是否适合你的服务、公司规模是否匹配你的目标客户、他们的购买意愿有多强等等。

Good lead might refer to how good a fit the industry is for your service, whether the company size matches who you like to serve, how strong their buying intent is, et cetera.

基本上，你需要把问题拆解成各个组成部分。

Basically, you're going to want to break questions into their constituent parts.

如果你的问题涉及评分，你需要用文字描述每个评分等级，而且你甚至可能想利用Jev的评分机会，多问一些你认为需要的问题。

If your question involves a scale, you need to describe each level of that scale in words, and you also might even want to take advantage of Jev scale opportunities to ask more questions than you think that you need.

最后，和所有AI一样，在实际投入生产之前，你需要做一些测试。

Lastly, like everything with AI, you're going to want to do some tests before you actually trust it in production.

这可能有点麻烦，但比如你用那个邮件分类器来按优先级整理东西，也许你可以自己给50封邮件打上标签作为测试，看看它的表现如何，确保它真的能按你的预期工作。

It might be a pain, but if you're using that email classifier, for example, to try to organize things based on priority, maybe you want to label 50 emails yourself as a test and see how it compares to make sure it's actually going to do for you what you want it to do.

现在，在我们结束之前，我会把这次演示文稿发布在本集的配套网站上，网址是aidailybrief.ai。最后几页会更实用一些，针对六个用例领域中的每个项目提供一个想法，比如内容档案分析器、研究评分器、收件箱分类器、改写检查器等等，而且里面还会有一个入门提示。

Now as we wrap up, I will be posting this presentation that I've been working through on this episode's companion site on aidailybrief.ai, and the last couple of pages get a little bit more practical with one idea for each project in the six use case areas, things like a content archive analyzer, a research rater, an inbox triager, a rewrite checker, et cetera, and there'll even be a starter prompt in there as well.

在Jev发布后的头10天里，我们很快就从热闹、令人兴奋的概念，发展到了真正有价值的实际应用场景。

In the first 10 days since Jev was released, we've gone from buzzy, exciting concept to actually valuable production use cases extremely quickly.

而且，因为这在本质上是一种新的基础能力，能够以极快的速度、几乎零成本地大规模应用简单判断，我认为我们需要一些时间才能真正弄清楚，我们能把它多么深入地融入到各种不同的使用场景中。

And yet because this is at core a new primitive in its ability to apply simple judgment at scale, at speed, and for effectively no money, I think it's going to take some time for us to really figure out just how deeply we can weave this into all sorts of different use cases.

随着越来越多的应用上线，我会回来分享其中最好的那些。

As more and more come online, I will come back and share the best of them.

不过，今天就到这里，这就是今天的AI每日简报。

For now, though, that is going to do it for today's AI Daily Brief.

一如既往，感谢你的收听或观看，下次再见，祝一切安好。

Appreciate you listening or watching, as always, and until next time, peace.