How to Decide What Work AI Should Do for You: The AI Deputization Audit — 文稿

M10:00

如果我告诉你，弄清楚你工作的哪些部分应该让AI来自动化，其实是一个简单的数学公式呢？

What if I told you that figuring out what parts of your work you should be getting AI to automate was a simple math equation?

这周，有两款新产品上线了，它们让AI帮你干活变得简单多了。

This week, 2 new products came online that make getting AI to do work for you much simpler.

GrokBot让用户通过手动录制操作来教它完成任务，而ChatGPT的电脑历史功能则观察你的工作方式，并随着时间推移不断学习。

GrokBot gives users the ability to teach it a task by manually recording them doing something, while ChatGPT's computer history watches how you work and learns over time.

这两者一起代表了AI最大的挑战从能力转向了背景。

Together, these represent the shift of the biggest challenge in AI moving from capability to context.

但随着这些新功能上线，你仍然需要弄清楚你工作的哪些部分想要让AI来自动化。

But as these new features come online, you still have to figure out which part of your work you want AI to automate.

最适合让AI代劳的工作特点是：频繁、耗时、可教、容易验证，并且不需要非得是你亲自做才能成功。

The work best suited for AI deputization is frequent, time-consuming, teachable, easily verifiable, and doesn't require you to have been the one to do it to be successful.

《AI每日简报》是一个每日播客和视频节目，涵盖AI领域最重要的新闻和讨论。

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

好了朋友们，在我们深入之前，先快速说几个通知。

Alright friends, quick announcements before we dive in.

M11:09

还有一件事，在我们进入头条新闻之前，包括Gemini的一个新模型。

And one more thing before we get into the headlines, including a new model from Gemini.

这周的很多节目，包括今天的节目，都与新发布的GrokBot有关。

A lot of the shows this week, including today's show, have to do with the newly launched GrokBot.

对于正在参加我们AI夏日冒险——也就是自选式冒险学习计划的朋友们，我们刚刚发布了一个新的弹出目的地，也就是一个项目，专门用来测试GrokBot。

And for those of you who are doing our AI Summer Adventure choose-your-own-adventure learning program, we've just posted a new pop-up destination, i.e., project, all about testing out GrokBot.

你可以在summeradventure.ai免费注册，然后了解如何使用这个永远在线的队友，我认为这是OpenAI的Claude有史以来最简单、最干净的版本。

You can sign up for free at summeradventure.ai and walk through how to use this always-on teammate that I think is the simplest, cleanest version of OpenAI's Claude we've ever had.

再说一遍，你可以在summeradventure.ai找到它。

Again, you can find that at summeradventure.ai.

但我们现在开始进入头条新闻。

But now let's get into the headlines.

感觉就像Google昨天听到了我们在谈论有多少人在讨论xAI，仿佛它已经完全取代了Google在严肃前沿模型公司中的地位。

It's almost like Google heard us yesterday talking about how many people were talking about xAI as though it had completely usurped Google in the pantheon of serious frontier model companies.

周四，Google发布了Gemini 3.7 Flash。

On Thursday, Google released Gemini 3.7 Flash.

虽然它既不是一再推迟的Gemini 3.5 Pro，也不是越来越受期待的Gemini 4，但它在效率方面开辟了一个不同的类别，而效率正变得越来越重要，尤其是对于严肃和高级用户来说。

And while it is neither the much-delayed Gemini 3.5 Pro nor the now increasingly anticipated Gemini 4, it does play in a different category of efficiency that's becoming a higher and higher consideration, especially for serious and advanced users.

那么我们先来看看这个模型的优点。

So let's start with what's good about this model.

它看起来非常非常快。

It appears to be very, very fast.

在Artificial Analysis的测试中，这个模型以每秒340个token的速度运行，这完全是另一个级别的表现。

During testing from Artificial Analysis, the model ran at 340 tokens per second, which is an entirely different category than anything else.

比GPT-4o快了两倍多，甚至比NVIDIA的新Nemotron 3.5 Lightning还要快一点。

It's more than twice as fast as GPT-4o and even a bit faster than NVIDIA's new Nemotron 3.5 Lightning.

在基准测试中，Google在3.6 Flash的基础上取得了一些扎实的提升，最显著的是在编程基准DeepSuite上的得分从48.6%提高到了65.3%。

On the benchmarks, Google made some solid gains over 3.6 Flash, most notably improving their score on coding benchmark DeepSuite from 48.6% to 65.3%.

而且这个模型，Google还将价格砍了一半，使得它比前代产品更具成本效益。

And with this model, Google is also slashing prices by half, making it a little more cost-effective than its predecessor.

不幸的是，和3.6 Flash一样，为了优化速度，这个模型最终陷入了一个奇怪的无人地带。

Unfortunately, just like 3.6 Flash, by optimizing for speed, the model kind of ends up in a strange no man's land.

即使成本降低了，在Artificial Analysis的基准测试中，这个模型每次任务仍然需要0.40美元。

Even with the cost reduction, the model still costs $0.40 per task on the artificial analysis benchmark run.

这使得它的价格和Mistral的Mixtral 1.2一样，比Nemotron 3 Ultra或GLM-4等模型略贵，大约是GPT-4o这类超便宜模型的8倍。

That makes it the same price as Mistral's Mixtral 1.2, slightly more expensive than models like Nemotron 3 Ultra or GLM-4, and around 8 times more expensive than the ultra-cheap models like GPT-4o.

与此同时，感觉这个模型在基准测试中的表现并不够强，难以证明其成本的差异是合理的。

At the same time, it feels like the model just isn't strong enough on the benchmarks to justify the cost difference.

Cognition指出，3.7 Flash的编码性能与Sonnet 5相当，但成本却不到一半。

Cognition pointed out that 3.7 Flash has Sonnet 5 coding performance for less than half the cost.

但说得客气点，Sonnet 5并不受欢迎。

But to put it mildly, Sonnet 5 has not been a hit.

大多数用户要么多花一点钱用前沿模型，要么在寻找一个更便宜的模型。

Most users are either paying a little more for a Frontier model or looking for a much cheaper model.

这倒不是说 Gemini 3.7 Flash 不好。

None of which is to say that Gemini 3.7 Flash is bad.

只是它目前在性价比曲线上的位置有点尴尬，不上不下的。

It just sits in an uncomfortable middle ground right now on the cost per intelligence trade-off.

但我认为，现在就断定我们完全了解这些模型的行为最终会如何演变，可能还为时过早。

And yet, I think it would be a mistake to assume yet that we really understand how on aggregate these behaviors are going to shape out.

实际上，有些用户报告说它的实用性提高了，特别是在编程能力提升方面。

In practice, some users are reporting increased utility, particularly with the boost in coding.

Vercel 的 Brandon Galang 写道：“别管那些对 Gemini 3.7 Flash 的负面说法。

Brandon Galang of Vercel wrote, "ignore the FUD on Gemini 3.7 Flash."

它其实是处于 Pareto 前沿的。

It actually sits on the Pareto frontier.

严格来说，它确实还是被 5.6 Luna 比下去了，但这很有迷惑性。

It still technically gets edged out by 5.6 Luna, but that's extremely deceptive.

5.6 Luna 的模型要小得多，而且虽然我的团队已经把它用在了很多生产工作流里，但我个人不会用它来做编程。

5.6 Luna is a much smaller model, and while my team has cut over a lot of production workflows to it, I personally would not turn to it for coding tasks.

Gemini 系列模型在多模态理解方面一直有很强的优势。

Gemini models have always had strong pros in multimodal understanding.

从模型行为的角度来看，我其实对 Grok 4.6 感觉有些不太放心。

I'm actually feeling somewhat iffy on Grok 4.6 from a model behavior standpoint.

所以目前，我打算把 Gemini 3.7 Flash 当作我的日常主力模型和执行模型来用。

So for now, I'm giving Gemini 3.7 Flash a try as my daily driver and execution model.

分析师 Max Weinbach 也同意，他说：“我在 Antigravity 上用了 Gemini 3.7 Flash，体验非常好。

Analyst Max Weinbach agreed, saying, "I'm really enjoying Gemini 3.7 Flash in Antigravity."

速度真的很快，效果也看起来非常不错，而且使用限制高得离谱。

It's really fast, seems to be really good, and the usage limits are insanely high.

有一点值得指出，对于部分同步的编程任务，也就是那些不是长期托管让 AI 自己干，而是你需要实时和它交互的任务，这种速度提升能带来很大的不同，也足以证明即使模型性能不如顶尖水平，稍微贵点也是值得的。

Now, it's worth pointing out that for partially synchronous coding tasks, i.e., coding tasks that aren't long horizon where you just go off and let the agent work, but where you are actually sitting there interacting with what the agent is producing, that speed boost could make a really big difference and actually justify a slightly increased price tag even at a decreased performance from the Frontier.

另一个重要的应用场景是改进 Gemini Spark，也就是谷歌的个人 AI 助手。

The other big use case is improving Gemini Spark, which is Google's personal agent.

巨大的速度提升、计算效率的提高，以及在编程和白领工作基准上的扎实改进，都使 Spark 得到了很大的升级。

The massive speed boost, the implied increase in compute efficiency, and solid improvements on coding and white-collar work benchmarks make it a big upgrade for Spark.

如果你要从这里面提炼出点什么，那就是又一个提醒：所谓的模型竞赛正在分化成多个不同的赛道。

If you're trying to take away anything from this, it's another reminder that the model race, quote unquote, is branching out into multiple races.

当然，还有争夺前沿模型的竞赛，但同时也存在争夺市场份额的竞赛。

There's certainly still the race for the frontier, but there's also a race for distribution.

还有围绕工程框架的竞赛，以及争夺收入的竞赛。

There's a race around harnesses, a race for revenue.

而且，单就模型本身而言，谷歌似乎正在押注速度也是一个人们会关注的维度。

And within the context of models alone, Google seems to be betting that speed is a dimension that people will pay attention to as well.

说到我们如何理解模型竞赛，我们刚刚收到一项非常有趣的新研究，它证实了我们在这个节目中经常讨论的一个观点：换用一个更便宜的模型，特别是中国模型，并不一定能给最终收益带来节省。

Now, speaking of how we understand the model race, we just got a really interesting new study that confirms something that we discuss on this show a lot, which is that switching to a cheaper model, particularly a Chinese model, won't necessarily deliver savings to the bottom line.

AlphaSense 的一项新研究考察了多种不同的美国和中国模型在实际任务中的表现。

A new study from AlphaSense looked at how a range of different US and Chinese models perform in real-world tasks.

它测试了 OpenAI 的 GPT-5.6 Sol，以及 Anthropic 的 Opus 4.8、Opus 5、Sonnet 5 和 Haiku 4.5。

It tested GPT-5.6 Sol from OpenAI, as well as Opus 4.8, Opus 5, Sonnet 5, and Haiku 4.5 from Anthropic.

来自中国的模型，AlphaSense 测试了 Kimi K3 和 GLM 5.2。

From China, AlphaSense tested Kimi K3 and GLM 5.2.

这项研究还包括两个美国开源权重模型：Thinking Machines Lab 的 Inkling 和谷歌的 Gemma 4。

The study also included 2 US open-weight models, Inkling from Thinking Machines Lab and Gemma 4 from Google.

这些模型被要求完成一系列几百道金融分析问题，需要它们筛选大量数据，包括财报电话会议记录、SEC 文件以及新闻文章。

The models were tasked with running through a series of several hundred financial analysis questions that required them to sift through a large volume of data, including earnings call transcripts, SEC filings, and news articles.

输出结果根据质量进行评分，评分标准包括事实准确性以及是否涵盖了多位分析师的观点等。

The outputs were scored for quality, which included things like factual accuracy and including multiple analyst perspectives.

结果显示，GPT-5.6 Sol 和 Opus-4.8 都能以比两个中国模型更低的成本提供更高质量的答案。

The results showed that both GPT-5.6 Sol and Opus-4.8 were able to deliver a much higher quality result for cheaper than either of the Chinese models.

例如，GPT-5.6 Sol 完成任务的成本比 Kimi K3 低了大约 13%，而质量评分却高出约 20%。

GPT-5.6 Sol, for example, completed the task for around 13% cheaper than Kimi K3, with a quality score around 20% higher.

Gemma 4 和 Inkling 也都表现不错，质量与 GLM 5.2 差不多，但成本却不到其五分之一。

Both Gemma 4 and Inkling were able to do a decent job, around the same quality as GLM 5.2, for less than 1/5 the cost.

Opus 5 是个有趣的例外，它的回复质量比 Opus 4.8 差，但成本却高出五倍以上。

Opus 5 was an interesting outlier, delivering lower quality responses than Opus 4.8 while also costing more than 5 times as much.

显然，这项研究的部分目的就是为了验证 Kimi K3 和 GLM 5.2 是否已经以极低的成本达到了前沿水平。

Now, clearly part of the purpose of the study was to test the theory that Kimi K3 and GLM 5.2 had reached frontier performance at a fraction of the cost.

实际上，研究发现，至少在这个用例上，这两个模型的表现和 Sonnet 5 差不多，但比 GPT-5.6 Sol 贵得多。

In practice, what the study found was that at least for this use case, they were both around the same level as Sonnet 5 and much more expensive than GPT-5.6 Sol.

AlphaSense 的 CEO Jack Coco 表示，有些更昂贵的模型——也就是那些只看每 token 价格显得更贵的模型——实际上最终成本更低，因为它们在 token 使用上更高效。

Said AlphaSense CEO Jack Coco, some of the more expensive models, the ones that look more expensive based on just their price per token, actually ended up being less costly because they were more efficient in using tokens.

特别是对各位企业买家和规划者来说，我们越快地转变对这个问题的传统认知，真正找出最适合我们自己特定任务的模型，你的预算状况就会越好。

Especially for you enterprise buyers and planners out there, the faster that we shift our conventional wisdom around this and actually figure out the models that are most efficient for our particular tasks, the better off your budget is going to be.

OpenAI 推出了 GPT-5.6 Sol 的超快模式。

OpenAI has introduced ultra-fast mode for GPT-5.6 Sol.

他们声称这种模式能以 14 倍的速度提供前沿智能。

Which they claim delivers frontier intelligence at 14 times the speed.

新模式让 Sol 的运行速度达到每秒 750 个 token，是 Gemini 3.7 Flash 的两倍多。

The new mode allows Sol to run at 750 tokens per second, which is more than twice the pace of Gemini 3.7 Flash.

目前，超快模式只对部分客户通过 API 开放。

For now, Ultrafast mode will only be available through the API to select customers.

这项功能利用了 OpenAI 为在 Cerebras 硬件上运行模型所做的优化，所以可用的基础设施会成为限制因素。

The feature leverages the optimizations OpenAI has made to run models on Cerebras hardware, so available infrastructure will be the limiting factor.

OpenAI 表示这只适合那些延迟是关键因素的特定工作流程，他们写道：超快模式专为那些更快的前沿智能能带来可衡量优势的企业设计，包括实时语音和客户支持、电商、编程与设计、金融研究以及安全响应。

OpenAI has positioned this as only really suitable for specific workflows where latency is a major factor, writing: Ultrafast is designed for businesses where faster frontier intelligence creates a measurable advantage, including real-time voice and customer support, commerce, coding and design, financial research, and security response.

公司没有透露超快模式的加价幅度，但估计不会太便宜。

The company did not mention how large the premium would be for Ultrafast mode, but presumably it's not ultra cheap.

不过，考虑到我们刚才提到的 Gemini 3.7 Flash，看到他们也进入这个领域还是挺有意思的。

Still, especially in the context of what we were just saying about Gemini 3.7 Flash, very interesting to see them push into this area as well.

还有一则来自 OpenAI 的消息，不过这次是关于人事的。

One other story from OpenAI, although this time on the personnel side.

周四，公司首席营收官 Denise Dresser 宣布，她将在未来几周内离开 OpenAI，去寻求其他机会。

On Thursday, the company's chief revenue officer, Denise Dresser, announced that she would be leaving OpenAI in the coming weeks to pursue other opportunities.

Denise 九个月前——也就是去年十二月——才加入公司，当时正值媒体质疑 OpenAI 增长势头的时候。她作为资深科技高管的丰富履历，在很大程度上缓解了投资者的担忧。

Denise joined the company just 9 months ago in December and was brought on at a time when OpenAI's growth was being questioned in the press, meaning that her significant pedigree as a veteran tech executive went a long way to easing investors' concerns.

在加入 OpenAI 之前，她在 Salesforce 工作了十多年，最后两年担任 Slack 的 CEO。

Prior to joining OpenAI, she had spent more than a decade at Salesforce, including spending her last 2 years as CEO of Slack.

与 OpenAI 的 CFO Sarah Friar 一样，Dresser 被视为新的业务导向高管团队的一员，目的是让公司在 IPO 前做好准备。

Alongside OpenAI CFO Sarah Friar, Dresser was viewed as a part of the new business-focused executive team designed to get the company in shape ahead of the IPO.

在宣布 Denise 离职的同时，公司也宣布了她的接替者，另一位行业资深人士——Wizz 的前总裁兼 COO Dolly Rajak。

At the same time as they announced Denise's leaving, the company also announced her replacement, this time with another industry veteran, Dolly Rajak, the former president and COO of Wizz.

总的来说，我不像其他媒体那样花太多时间在这些人事变动上。

Now, in general, I don't spend nearly as much time on these sort of personnel moves as some other media properties do.

我认为十次里有九次，对于人事变动的奥卡姆剃刀解释其实就是个人原因。

I think that 9 times out of 10, the Occam's razor explanation for personnel moves is in fact personal.

我也觉得，外界观察者太想从任何变动中解读出重大问题的迹象，所以试图从外部做大量分析相当困难。

I also think that there is such a desire right now among outside observers to read any sort of shifts as tea leaves for major problems inside that trying to do a lot of analysis from outside is pretty fraught.

不过，抛开我个人的看法，市场显然注意到了这次变动。

However, holding aside whatever I think, the market is certainly noticing this move.

部分原因是，这距离前 COO Brad Lightcap 的离职仅隔了几天；另一部分原因是，这两次离职加上过去几个月里一系列的人事出走，看起来形成了一种模式。

Part of that is that it follows by just a couple of days former COO Brad Lightcap's leaving, and part of that is that those 2 departures form what looks like a bit of a pattern running for the exits over the past couple of months.

在前应用与 AGI 部署负责人 Fiji Simo 七月份因健康问题离职后，据 Axios 的消息人士透露，总裁 Greg Brockm an 一直在组建自己的领导团队，而这次变动可能也是其中的一部分。

After former CEO of Apps and then AGI deployment Fiji Simo left in July due to health concerns, sources speaking to Axios suggest that President Greg Brockman has been building up his own team of leaders, and this move may be part of that.

无论如何，在 IPO 之前，高管离职被认为是一个问题。

Whatever the case, executive turnover is being flagged as a problem prior to the IPO.

不过，既然 IPO 已经推迟到明年，OpenAI 有充足的时间在登陆华尔街之前构建另一种叙事。

But then again, with that now delayed until next year, OpenAI has a lot of time to build another narrative before their Wall Street debut.

这部分我们就先讲到这里，今天的头条新闻就结束了。

For now, we will leave that there, and that's going to do it for today's headlines.

接下来，是主节目。

Next up, the main episode.

M113:30

欢迎回到《AI 每日简报》。

Welcome back to the AI Daily Brief.

今天我们做一期节目，这种类型以后我会尝试更多，就是把一个新闻故事和一个小活动结合起来，希望能让你觉得实用。

Today we're doing a type of episode that I'm going to be trying out a lot more, which is effectively combining a news story with a small activity that hopefully makes this practical for you.

今天的活动叫做‘AI委任审计’，是一个简短的过程，帮你弄清楚你能把什么、应该把什么交给AI来做。

The activity today is called the AI Deputization Audit, and it's a short process for helping you figure out what you can and what you should hand over to AI.

启发这个活动的新闻故事其实有两部分。

Now, the news stories that inspired this are actually two-part.

本周我们看到了两个新功能的发布，它们反映了AI的瓶颈已经从模型能力转移到了上下文获取上。也就是说，模型能做某事，并不意味着它拥有足够的信息和上下文来为你个人做好这件事。

This week we got 2 new features released that reflect the fact that the bottleneck in AI has moved from model capability to access to context, i.e., just because a model can do something doesn't mean it has the information and context it needs to do it well relative to you personally.

这个问题已经存在很久了，我们也尝试了很多不同的实验来改善它。

Now, this has been a problem for a very long time, and we've explored a bunch of different experiments to improve that.

比如，你仍然可以去contextportfolio.ai，和我几个月前设置的一个AI进行对话，来构建一个可迁移的个人上下文文件。

For example, you can still go to contextportfolio.ai and do an interview with an AI that I set up a few months ago to build a transportable personal context file.

不过，跟本周发布的新东西比起来，这可能显得特别慢了。

That, however, might seem incredibly slow given what has launched this week.

本周发布的第一个此类功能是GrokBot的一个新能力。

The first feature in this area that launched this week was one of the new capabilities of GrokBot.

如果你错过了那期节目，说明一下：GrokBot是Cursor和SpaceX AI新推出的简化版OpenCLAW，但面向的是普通用户。

For those who missed that episode, GrokBot is Cursor and SpaceX AI's new simplified version of OpenCLAW, but for everyone else.

其中最大、最有趣的功能之一，就是你可以教GrokBot完成一个任务。

And one of the biggest and most interesting features is the feature by which you can teach GrokBot a task.

你只需要在聊天界面点击一个小加号按钮，然后录下自己在浏览器中操作的过程。

You simply hit a little plus button in the chat and record yourself doing something in the browser.

机器人看着你的操作，理论上之后它就能重复执行了。

The bot watches and then theoretically it can do again.

虽然Cursor团队说GrokBot已经能自己做很多事了，但如果你发现它不太行，教它一个任务就能一举解决问题。

And while the team at Cursor say that many things GrokBot can already do by itself, if you ever find your bot struggling, teaching it a task is a way to solve that in one fell swoop.

这个功能本身已经很有意思了，但更妙的是，我们还看到了另一个同类功能，这次来自OpenAI。

Now that feature was interesting all on its own, but it was made more so by the fact that we got another feature in the same family, this time from OpenAI.

另一个是ChatGPT的功能，叫‘计算机历史’（Computer History）。

That other feature is for ChatGPT and is called Computer History.

OpenAI的Ari Weinstein写道：‘计算机历史让ChatGPT从你在电脑上做的一切中学习，从而更好地理解你的工作方式，完成你正在进行的任务，并根据你的使用习惯推荐技能和自动化方案。’

OpenAI's Ari Weinstein writes, computer history lets ChatGPT learn from everything you do on a computer so it can better understand how you work, finish tasks that you're in the middle of, and suggest skills and automations based on how you use your computer.

它的目标和教GrokBot完成任务完全一样。

The goal is exactly the same as teaching GrokBot a task.

就是向ChatGPT展示你的工作方式，这样它就能帮你做更多的工作。

It's to show ChatGPT how you work so that it can do more of that work.

现在，一些细心的听众可能会想，等等，我们是不是之前听说过类似的功能？

Now, some of you careful listeners might be thinking to yourselves, wait a second, didn't we hear about some feature like this?

不久之前还引起过不小的争议？

Not so long ago that caused a bunch of controversy?

果然，X平台上有像Shuv这样的用户说：‘我年纪够大了，还记得微软用Windows Recall做过同样的事，当时大家都疯了。’

And sure enough, X users like Shuv said, I'm old enough to remember when Microsoft tried to do the same thing with Windows Recall and everyone lost their minds.

Shuv指的是微软在2024年初推出Copilot+ PC时宣布的一个功能。

What Shuv is referring to was a feature that was announced around the time that Microsoft introduced its Copilot+ PCs back in early 2024.

《时代》杂志这样解释这个功能：‘Recall让设备每隔几秒就截取一次用户的屏幕快照。’

Time explained the feature like this: Recall allows the device to take snapshots of a person's screen every few seconds.

这些快照经过加密，然后存储在用户设备的本地。

These snapshots are encrypted and then stored locally on the individual's device.

微软表示，这个功能的设计目的是‘解决我们每天遇到的最烦人的问题之一：在电脑上找到我们之前见过的东西。’

Microsoft said that this feature was designed to, quote, solve one of the most frustrating problems we encounter daily, finding something we know we have seen before on our PC.

但人们并不喜欢这个功能。

But people did not like this.

Chris Srisak博士告诉BBC：‘这可能是一场隐私噩梦。’

Dr. Chris Srisak told the BBC, this could be a privacy nightmare.

仅仅是使用设备时会被截屏这一事实，就可能对人们产生寒蝉效应。

The mere fact that screenshots will be taken during use of the device could have a chilling effect on people.

最终，微软召回了Recall，后来又发布了一个版本，做了很多改动，包括改为用户主动选择加入，并给予用户更精细的控制权限。

Now ultimately, Microsoft recalled Recall and did later release a version making a bunch of changes, including making it opt-in and giving users a lot more fine-grained control.

那么，是什么改变了呢？

So what has changed?

这只是时间问题吗？

Is it just a matter of time?

Javier LaCour 写道，2022 年，我不信任 Google。

Javier LaCour writes, 2022, I don't trust Google.

我会习惯在手机上用无痕模式。

I'm going to get used to using incognito mode on my phone.

我换成了 Firefox，VPN 一直开着。

I switched to Firefox, VPN always on.

2026 年，拿走我的浏览记录、我的医疗数据、我的银行流水、访问我的 Gmail、分析这段 2018 年的 WhatsApp 聊天记录。

2026, Take my history, my medical analytics, my bank statements, access to my Gmail, analyze this WhatsApp conversation from 2018.

AI 创业者和内容创作者 Theo 写道，我深陷 AI 中毒太深，居然觉得这听起来很棒。

AI entrepreneur and content creator Theo writes, I'm so deep in my AI psychosis that I think this sounds great.

说实话，从 Windows Recall 或 Rewind 公布那会儿我就想要这个了。

I'm not gonna lie, I wanted this since Windows Recall or Rewind or whatever was announced.

当时被骂得太惨了，我就没敢吭声。

It got panned so hard that I kept my mouth shut.

过了这么久，我决定豁出去了，试试这个。

Enough time has passed that I'm gonna say screw it and try this.

我得说，我觉得这有一部分只是人们对隐私的态度在发生变化。

And I will say that I do think part of it is simply shifting attitudes on privacy.

如果我没记错的话，我当时对 Recall 的评论是，这听起来就像那种现在的人会觉得完全疯狂，但未来的人会惊讶我们以前居然没有的功能。

If I recall correctly, my comments on Recall at the time were that it sounded to me exactly like the type of feature that people now would think was completely insane and that people in the future couldn't believe we previously didn't have.

但与此同时，价值主张上明显有区别。

At the same time though, there is very clearly a difference in the value proposition.

想想我刚才说的那句话。

Think about that sentence I said.

微软把这个功能宣传成解决“在电脑上找到我们以前见过的东西”的问题，但这真的是那么大的问题吗？

With Microsoft pitching this as solving the problem of finding something we know we've seen before on the PC, is that really that big a problem?

它大到值得去冒所有隐私风险吗？

Is it a big enough problem to be willing to risk all the privacy considerations?

我觉得对很多人来说并不是。

I don't think for many people it is.

但让一个 AI 代理真正替你干活，那价值主张就完全不同了。

But getting an AI agent to actually do your work for you is a much different value proposition.

很明显，我们现在生活在一个世界里，这些代理获取信息来实际完成工作，有了更多选择。

And what's clear is we now live in a world where there's a lot more choice on how those agents get the information they need to actually do the work.

比如 Simon Smith 指出，computer history 克服了之前尝试的局限，因为它不是不停地截图，而是记录交互事件，不是屏幕或音频。

Simon Smith, for example, points out that computer history overcomes a limitation of previous attempts because instead of taking screenshots constantly, it's recording interaction events, not screen or audio.

大体来说，ChatGPT 的 computer history 和 GrokBot 的 teach-a-task 代表了 AI 学习你工作方式的两种不同途径。

Broadly speaking, ChatGPT's computer history and GrokBot's teach-a-task reflect 2 different ways that AI can learn how you work.

第一种，以 computer history 为代表，我们可以叫它被动观察。

One, embodied by computer history, we might call ambient observation.

也就是 AI 跨应用观察，不断积累上下文，在你做事的时候学习你做什么，你本人不需要特别费心。

This is where the AI watches across apps, builds ongoing context, and can learn about what you do as you do it without any particularly strong consideration on your part.

而 GrokBot 的 teach a task 模式，我们可以叫它刻意演示。

The GrokBot teach a task paradigm we might call deliberate demonstration.

也就是你主动按一个按钮，明确告诉它，我现在要教你一项技能。

This is where you intentionally press a button where you are deciding and telling it, I am going to now teach you a skill.

你按下 teach a task，展示工作流程，它会把这些步骤存为一个例程，以后就能重复这个流程。

You press teach a task, show it the workflow, it saves the steps as a routine, and it can repeat that process later.

这两种模式各有各的好处。

Both of these paradigms have things to recommend them.

对于 computer history 来说，很多人觉得它就在后台默默运行，这个特点本身就很吸引人。

On the computer history side, for a lot of folks, the fact that this is just happening in the background as they work will be the killer feature.

他们不需要做任何有意识的操作，而它却依然在学习如何变得更实用，这本身就是很大的价值。

The fact that they don't have to do anything intentional, and yet it's still learning how to be more useful, is a big part of the value.

另一方面，我觉得很多人可能更偏好 GrokBot 那种刻意演示的方式。

On the other hand, I can see for a lot of folks the deliberate demonstration approach of GrokBot being much preferable.

对有些人来说——我大概也属于这一类——更自然、更喜欢的模式是，主动决定哪些事情要演示给代理看，这样它就能替你做这些事。

For some, and I would probably put myself in this category, it is likely to be a much more natural and preferred pattern to be intentional about which things you want to actually show the agent how you do so that it can presumably do them for you.

但当然，问题就变成了：你应该把工作中的哪些部分委派给AI？

But then of course the question becomes, Which parts of your work should you be deputizing to AI?

不可避免的是，不同的人会对这个问题有不同的答案，但为了帮助你思考你工作清单中的哪些部分可能适合这些新能力，让我介绍一下委派审计。

Inevitably, different people are going to come up with different answers to this question, but to help you think through which parts of your work catalog might be well suited to these new capabilities, let me introduce the deputization audit.

你会注意到，我没有使用“自动化”这个词，这是故意的。

You'll notice that I am not using the word automation, and I'm doing that very intentionally.

对我来说，至少个人而言，把AI委派去做某件事的概念，感觉更像是我想要与AI建立的关系，而不是自动化掉然后永远不再想那个任务。

For me, at least personally, the concept of deputizing AI to go do something in my stead feels a little bit more like the relationship that I want to have with AI, than automating away and never thinking about some task again.

我会把这份委派审计作为节目的延伸发布出来。

I will publish this deputization audit as an extension of the show.

所以如果你访问aidailybrief.ai并点击今天的剧集，你会找到链接。

So if you go to aidailybrief.ai and click on today's episode, you will find a link to it there.

第一步是盘点你重复出现的流程，这样我们才能用这个系统的其余部分来评估它们。

Step 1 is to take an inventory of your recurring processes so that we can run them through the rest of this system.

我在这里专注于重复出现的，因为一般来说，这类工作流程最适合这种自动化或委派。

I'm focused here on recurring because in general, those are the types of workflows that this sort of automation or deputization is going to be well suited for.

而且总的来说，我发现无论我们多么努力让工作变得新颖、动态和不同，很多工作其实就是这些我们日复一日或周复一周必须要做的事情。

And in general, I've found that no matter how much we try for our work to be novel and dynamic and different, A lot of it is just this stuff that we have to do day in, day out or week in, week out.

所以想想像邮件分类、每周状态报告、会议准备、研究简报、CRM或销售管道维护、内容再利用、日程安排和差旅、供应商门户杂务、潜在客户资格认定、指标和分析数据拉取这些。

So think about things like email triage, weekly status reports, meeting prep, research briefs, CRM or pipeline hygiene, content repurposing, scheduling and travel, vendor portal chores, inbound lead qualification, metrics and analytics pulls.

所有这些都是即使是最有活力的人也可能发现自己要反复应付的事情。

All of these are things that even the most dynamic person might find themselves slogging through a recurring process for.

第二步是，对于这些不同的流程，我们要给它打分，分数会跨越5个不同维度。

Step 2 is that for each of those different processes, we're going to give it a score, and the score is going to be across 5 different dimensions.

总的来说，这就是我提出的5个维度，它们决定一个任务是否适合自动化或委派。

Overall, then, these are the 5 dimensions that I'm proposing make a task well-suited to automation or deputization.

第一个是这个任务是否值得自动化。

The first is whether the task is even worth it to automate.

第二个是这个任务是否可教。

The second is whether the task is teachable.

第三个是这个任务或流程是否可检查。

The third is whether the task or process is checkable.

第四个是任务的赌注是什么，尤其是当出错的时候。

The fourth is what the stakes of the task are, especially if something goes wrong.

第五个标准是，你在多大程度上是输出质量的关键因素。

And the fifth criteria is how much you personally are the key factor in the quality of the output.

换句话说，别人来做这件事，是不是也一样有用、一样有价值，或者一样能达成目标？

In other words, could someone else be doing this and it would be just as useful or just as valuable or accomplish its goal just as well?

所以标准一是某个流程是否值得自动化或委派给AI去做。

So criteria 1 is whether a process is even worth it to automate or deputize AI to do.

那基本上归结为你多久做一次这件事，以及需要花多少时间。

And that basically comes down to how often you do this thing and how much time does it take.

0分意味着这件事不常发生，而且即使发生了，也不需要花很长时间。

A score of 0 means that it doesn't happen very often and that it doesn't take very long to do when it does happen.

想想一年几次，一次几分钟。

Think a few times a year and a few minutes at a time.

1分是它发生的频率适中，需要一些时间但不是太多时间。

A score of 1 is that it happens moderately frequently, and takes some time but not a ton of time.

比如说，某个流程大多数星期都会发生，而且在一个小时内完成。

So, for example, some process that happens most weeks and is done in under an hour.

2分是流程非常频繁地发生，而且需要大量时间。

A score of 2 is a process that happens very frequently and takes a lot of time.

例如，每周或更频繁，而且持续数小时。

For example, weekly or more and extending into the hours.

显然，你可以看到，分数越高，这个任务就越适合至少去探索AI委派。

Obviously, you can see here that the higher the score, the more well-suited to at least explore AI deputization that task might be.

第二个标准是可教性。

Second criteria is teachability.

你能用10分钟的屏幕分享来展示这个任务吗？

Could you show this task in a 10-minute screen share?

0分基本上就是完全不行。

A score of 0 is basically absolutely not.

要花好几个月才能教会，或者根本就没法用这种方式教会。

It would take months to teach or is just impossible to teach in this way.

1分的意思是你能演示一下，但会有很多限制条件。

A score of 1 is that you could demo it, but there would be a lot of caveats.

2分的意思是只需要一次演示就能轻松覆盖，并且教会AI那个技能。

A score of 2 means one demo could easily cover it and teach the AI that skill.

第三个标准是关于输出结果的可检查性。

The third criteria is about the output results, the checkability.

验证输出所需要的时间，和你自己动手做出来所需的时间相比，哪个更短？

How long does it take to verify the output compared to how long it would take to produce that output?

显然，如果检查时间跟自己做一样长，那自动化或者AI代劳就省不了你时间了。

Obviously, if checking takes as long as doing it yourself, then automation or AI deputization is not going to be saving you time.

所以在这种情况下，0分就表示要检查就等于要重做一遍。

So in this case, a score of 0 reflects the idea that to check it means redoing it.

1分的意思是它可检查，但你需要仔细读一遍。

A score of 1 means that it is checkable, but you're going to have to give it a careful read.

2分的意思是扫一眼就能知道所有你需要的信息。

A score of 2 means that a quick glance can tell you all you need to know.

标准4是风险程度，也就是说，如果AI搞错了而且没人发现，后果有多严重？

Criteria 4, we have the stakes, i.e., how bad is it if the AI gets it wrong and nobody catches it?

0分——再次说明，这表示不太适合自动化——实际上就是风险很高，如果AI搞错了，问题会很严重或者不可逆。

A score of 0, which again means that it's less likely to be well-suited for automation, is effectively that the stakes are high, that the problems would be serious or irreversible if AI got it wrong.

1分表示出问题会让人难堪但还能补救，2分表示风险很低，出了错也很容易重来。

A score of 1 means that doing something bad would be embarrassing but fixable, and a score of 2 means that it's pretty low stakes, and if something goes wrong, it's easy to redo.

最后一个标准是你在流程中的不可或缺程度。

Last criteria is how integral to the process you are.

0分表示必须得是你，这又说明这个流程非常不适合AI代劳。

A score of 0 means it has to be you, again making that particular process pretty ill-suited for AI deputization.

1分表示有你参与有帮助，但不是必须的；2分表示只要活儿干完了，根本没人会注意到是你做的还是外面一只松鼠做的。

A score of 1 means that your involvement helps but isn't essential, and a score of 2 means that literally no one would notice if you did it or if a squirrel from outside did it as long as the work got done.

那么这样算下来，满分10分，你有一个总分。

Now from there, you have a combined score across the 10 total points.

我建议分成三个档次。

I propose breaking it into 3 tiers.

8到10分表示非常值得考虑让AI来代劳这个任务。

A score of 8 to 10 means that it's very worth considering deputizing AI to do that task.

8到10分意味着风险很低，任务发生频率很高，很容易教会，而且输出结果是你做还是别人做都无所谓。

A score of 8 to 10 means the stakes are fairly low, the task happens a lot, it's highly teachable, and the output doesn't really matter whether it's you or someone else.

这类事情你就应该交出去，抽查一下结果，看看做得怎么样。

Those are the types of things that you're going to want to hand over, spot check the output, and see how it does.

而且如果你在尝试类似Computer Use或GrokBot的Teach a Task这类工具，这个区间内的流程可能就是最值得先试试的。

And if and when you are experimenting with something like Computer Use or GrokBot's Teach a Task, processes that fall in this bucket are probably where you want to look first.

而在另一个极端，如果分数在0到3之间，我们不建议代劳，而是建议“守护”，也就是说你应该自己保留这项工作。

Now, on the other end of the spectrum, when the score is between 0 and 3, instead of a suggestion of deputizing, we'll call that a defend, i.e., that you should keep that work to yourself.

可能是因为犯错代价太高。

Maybe this is because a mistake would be expensive.

也可能是因为电话那头的人期望是你本人接。

Maybe it's because the person on the other end of the line expects this to be you.

很有可能就算不打分，你也能判断哪些任务需要守护，但至少这个标准让判断更一致了。

It's highly likely that even without a score, you will understand which tasks you need to defend, but at least this puts some consistency around it.

当然难点在于，很多工作会正好落在中间区域。

The challenge will, of course, be that a lot of stuff is going to fall right in the middle.

4到7分，我称之为“二重奏”，就是AI做一部分，但你仍然高度参与。

A score of 4 to 7, I'm calling a duet, where you're going to have AI do part of the task, but you're still going to be highly involved.

坦白讲，今天大多数知识工作都落在这个类别，问题在于随着时间的推移，AI和学习你工作流程的方法是否能变得足够好，让更多“二重奏”任务转变为“代劳”任务。

Frankly, most knowledge work today is going to sit in this category, and the question will be over time whether AI and the process by which it learns what you do gets better enough that more of those duet tasks can move and become deputized tasks instead.

这就引出了第4步：明确障碍。

Which brings us to step 4, which is naming the blocker.

对于落在类似二重奏或防守类别的那些任务或流程，具体是什么会阻止你把这些交出去？

What would specifically prevent you from handing this off for those tasks or processes that land in something like the duet or defend category?

所以，一些常见的阻碍因素可能是像工作跨网站或旧软件进行的情况。

So some common blockers might be things like the work happens across websites or legacy software.

想想那些没有 API 的供应商门户网站。

Think vendor portals that have no API.

另一个阻碍可能是流程很难用单一提示来解释，也就是说，你知道怎么做，但很难写下来。

Another blocker might be that the process is hard to explain in a single prompt, i.e., you know how to do it but would struggle to write it down.

这两类就是阻碍因素。

These are two categories of blockers.

识别具体阻碍因素之所以重要，是因为我们手中工具的新更新可能会改变某些阻碍的解决状况。

The reason that identifying the specific blocker matters is that new updates to the tools that we have might change the equation around certain blockers.

比如说，对于我刚才提到的这两种，这些新系统，Computer Use 和 GrokBot，都有可能改变并解决这些阻碍。

For example, for the 2 that I just mentioned, these new systems, Computer Use and GrokBot, both potentially change and solve those blockers.

对于没有 API 的系统，能像你一样点击同一屏幕的 computer use 代理，意味着这可能不再是问题。

For systems with no API, computer use agents that click the same screens you do means that that might no longer be a problem.

而对于那些容易演示但难以解释的流程，嗯，你可以直接展示而不是描述。

And for processes that are easy to show but hard to explain, well, you can just show rather than tell.

其他一些阻碍可能会被这些新工具解决，但不太明显。

Some other blockers might be solved by these new tools but aren't quite as clear.

比如说，AI 缺乏必要的上下文，也就是不知道你的账户、你的历史记录或你的格式。

For example, the AI lacking necessary context, i.e., not knowing your accounts, your history, or your formats.

像 computer use 这样的东西可能会解决这个问题，通过持续观察可以逐步积累，但用 GrokBot 花十分钟教一个任务可能做不到。

Something like computer use might solve that, where ongoing observation over time could build that up, but a 10-minute teach-a-task with GrokBot probably won't.

现在，其他一些阻碍不会因为这些新工具而得到解决。

Now, some other blockers are not going to be solved by these new tools.

比如说，工作需要品味或判断力，错误代价高昂或不可逆，或者工作依赖于人际关系。

For example, the work requiring taste or judgment, mistakes being costly or irreversible, or the work depending on human relationships.

所有这些都不会因为这种新的教学任务能力而天然得到解决。

None of those things are inherently solved by these new sort of teach task capabilities.

而且值得注意的是，和其他事物一样，新能力有时确实会带来新挑战。

And it is also worth observing that as with anything, new capabilities do sometimes create new challenges.

如果 AI 委派的阻碍是隐私或安全问题，让 AI 做这个不合适，那么这些新工具实际上可能增加阻碍，因为录屏会产生新的安全事项。

If the blocker to AI deputization is that privacy or security makes it inappropriate for the AI to do that, these new tools could actually in fact add a blocker as recording your screen creates a new thing to secure.

不过，第 4 步是在你识别出 Duet 和 Defend 中任务的阻碍之后，看看这些阻碍是否有任何已被这些新工具解决。

Still, step 4 is that after you have identified the blockers to your tasks in Duet and Defend is to see whether any of those blockers are now solved by these new tools.

说到这，你现在就有了一个委派清单，也希望有一些新想法，关于如何利用这些新工具。

And with that, you now have a deputization list and hopefully some new ideas for how you can take advantage of these new tools.

现在，关于第 2 天和第 3 天对 GrokBot 的反应，我在上一集中非常热情地谈到它，更多人体验它确实意味着在某些情况下人们开始遇到它的局限。

Now, when it comes to day 2 and day 3 sort of responses to GrokBot, which I spoke about very glowingly on the episode before, more people experiencing it does mean in some cases that people are coming up against its limits.

比如说，Super Trainer，我这么说是双关，既指她很出色，也指她和超级智能的 Nufar Gaspar 合作，她发现虽然她觉得 GrokBot 对很多还没建立完整代理系统的人很有用，但对于更高级的用户，有很多挑战。

For example, Super Trainer, and I mean that both in that she is super and that she works with super intelligent Nufar Gaspar, found that while she felt like GrokBot would be good for a lot of folks who hadn't built complete agent systems yet, For more advanced users, there were lots of challenges.

她不喜欢的是，不能更具体地指定哪个文件夹包含相关上下文，而且在模型选择等方面没有控制权。

She didn't like that she couldn't be more specific about which folder holds the relevant context and didn't have control on things like model choice.

所以，尤其是对于一些更高级的用户，那些上过 ClockCamp 的人，GrokBot 可能最终不是完美的选择。

So especially for some of you more advanced users, the type of folks who have gone through ClockCamp, GrokBot might not ultimately be a perfect fit.

我们开始听到一些人们从中获得价值的使用案例。

And we're starting to hear some of the use cases that people are getting value from.

John O'Neill，他拥有一家管道公司，在 24 小时内从零开始实现了自动调度和办公室杂务，尽管他没有雇佣任何工程师。

John O'Neill, who owns a plumbing company, went from zero to automated dispatch and office chores in 24 hours, despite not having any engineers on the payroll.

很多人正在使用 Chief of Staff 功能，他们为单个任务创建单独的机器人，但只通过他们的主 Chief of Staff 进行交互，这个主机器人协调所有其他机器人。

Lots of people are using the Chief of Staff capabilities where they spin up individual bots for individual tasks, but then only interact through their main Chief of Staff who coordinates all the other bots.

确实，每个机器人对应一个话题的模式似乎是早期出现的最佳实践之一。

Indeed, that pattern of topic per bot seems to be one of the early emerging best practices.

最后，我经常看到的一个用例，可能值得你实验的是收件箱、Slack 追踪和早晨简报。

Finally, one use case that I'm seeing a lot that might be a good one to experiment with is the inbox, Slack tracker, and morning brief.

Matt Van Horn 称这是通用入门任务，和以前每个代理一样，只不过现在设置大约需要一分钟。

Matt Van Horn calls this the universal starter job, same as it was for every agent before it, except now it takes about a minute to set up.

无论你测试什么，如果你能访问 GrokBot 或 Computer History，我确实认为值得花一些认真时间，实验这些新型 AI 委派工具能为你做什么。

Regardless of what you test, if you have access to GrokBot or Computer History, I do think it's worth spending some conscientious time experimenting with what these new sort of AI deputization tools can do for you.

即使对高度熟悉AI的人来说，最大的挑战之一也是挤出时间来学习用不同的方式做事。

One of the biggest challenges, even for people who are highly AI-enfranchised, is carving out the time to learn to do something differently.

事实上，尤其是那些高效高产的人，往往把事情安排得井井有条，以至于花点时间彻底改变做事方式，短期内感觉像是浪费时间，即使从长远来看能省下不少时间。

In fact, especially high-efficiency and high-productivity workers often have the way that we do things so dialed in that taking a little while to totally change that feels like a waste of time in the short term, even if it would save us a lot of time in the long term.

不过，我觉得这些实验还是值得的。

Still, I think the experiments are worth it.

所以，对于想试试的朋友，留意一下GrokBot在扩大服务时成本的下降，然后告诉我你的发现。

And so for those of you who want to try, keep an eye on the cost coming down on GrokBot as they expand access, and tell me what you find.

今天就到这里，以上就是今天的AI每日简报。

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

一如既往，感谢你的收听或观看。

Appreciate you listening or watching as always.

下次再见，祝一切安好。

And until next time, peace.

已剔除 2 处广告（点击展开查看）

M10:54广告 · 已剔除

First of all, thank you to today's sponsors, KPMG.

Blitzy, Harbor, and HyperAgent.

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

To learn more about sponsoring the show, send us a note at [email protected].

M110:29广告 · 已剔除

If you're leading AI inside an enterprise, you already know that the gap right now isn't capability but execution.

That's why KPMG's You Can with AI is back with a new season featuring conversations with leaders like Surojit Chatterjee of EMA, Mei Habib of Ryder, McKesson CIO Ellery Fisher, and others focused on practical execution.

What's working, what's not, and what it actually takes to move from pilots to real scaled impact across strategy, data readiness, governance, workforce, and value.

And of course, it's co-hosted by me, Nathaniel Whittemore.

Go listen and subscribe at www.kpmg.us/aipodcasts.

That's www.kpmg.us/aipodcasts.

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

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

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

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

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.

One Blitzy customer stood up a brand new application— 534,000 lines of code— compressing a 65-week roadmap into 2 weeks.

Another shipped an entire application with no frontend engineer.

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

Build what's next at blitzy.com.

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

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

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' AI Lab Ecosystem ETF suite gives investors a way to gain exposure to the AI ecosystem they believe is best positioned for success.

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

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

Read and consider it carefully before investing.

Risks include principal loss and artificial intelligence-related risks.

Harbor ETFs are distributed by Foresight Fund Services, LLC.

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

This is a paid advertisement and not personalized investment advice.

Investing involves risk, including possible loss of principal.

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

New users get $1,000 in inference.

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

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

Marketing's agent turns competitor moves into landing pages.

Sales' agent enriches leads, drafts emails, and updates the CRM.

Ops' agent chases the paperwork and tracks the budget.

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

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

Hire yours at HyperAgent, built by the team at Airtable.

Claim your $1,000 in inference at hyperagent.com/AIDailyBrief.