#238：一家700人规模的银行如何利用AI构建应用程序、智能代理和数字员工

如果我们希望AI像我们期待的那样真正成为团队的延伸，我们就需要改变与AI互动的方式。

And if we want AI to be a true extension of our teams like we had hoped, we need to change how we interact with AI.

这就是我们如何接受数字员工这个概念的过程。

And that's how we came to embrace this concept of of digital employees.

欢迎收听《AI转型》，这是《人工智能秀》的一个特别系列。

Welcome to AI Transformations, a special series from the Artificial Intelligence Show.

我是Mike Kaput，SmarterX和营销AI研究所的首席内容官，也是你们的主持人。

I'm Mike Kaput, Chief Content Officer at SmarterX and Marketing AI Institute, and I'll be your host.

每个企业的AI之旅都各不相同。

Every business's AI journey looks different.

在每一集中，我都会与那些经历过真实AI转型的领导者对话，包括真实的故事、促使他们行动的痛苦点、事情开始顺利的时刻，以及他们今天能展示的成果。

In each of these episodes, I sit down with leaders who have lived through real AI transformation, including the actual stories, the pain points that push them to act, the moment things started to click, and the results they can point to today.

所以，加入我们，一起在《AI转型》系列中提升所有人的AI素养。

So join us as we accelerate AI literacy for all as part of our AI Transformations series.

欢迎大家收听《人工智能秀》第238集。

Welcome everyone to episode two thirty eight of the Artificial Intelligence Show.

我是迈克·卡普特，人工智能秀的联合主持人，也是SmarterX的首席内容官。

I am Mike Kaput, cohost of the Artificial Intelligence Show and chief content officer here at SmarterX.

再次说明，今天的节目和我们平时的内容有点不一样。

Once again, today's episode is a little different than our regularly scheduled programming.

这是一个特别节目，属于我们正在推出的一个系列，叫“AI转型”，由我们的朋友谷歌云赞助。

This is a special episode in a limited series we're running called AI Transformations presented by our friends at Google Cloud.

在这个系列里，我们聚焦的是真实公司如何用AI推动真正的变革。

So in this series, we are spotlighting how real companies are driving real change using AI.

在每一集中，我们会探讨全球一些最具创新力的公司的领导者，如何积极利用AI来改变他们的团队、部门甚至整个组织的工作方式。

So in each of these episodes, we're going to explore how leaders at some of the world's most innovative companies are actively using AI to transform how their teams, departments, or even entire organizations work.

我们会看看是什么触发了转型，AI之前的工作是什么样的，转型过程是如何展开的，中间遇到了哪些麻烦，以及AI现在开始为企业带来哪些成果。

We're going to look at what sparked transformation, what work looked like before AI, how journeys unfolded, what got messy along the way, and what results AI is starting to unlock for businesses.

我们的方式是直接在节目里采访这些领导者，包括今天和我一起的约翰·科瓦尔，来自PPAC私人银行与信托。

And the way we do this is we interview leaders firsthand right here on the show, including John Kowal from PPAC Private Bank and Trust, who is here with me today.

关于约翰和他的工作，稍后会有更多内容。

Much more on John and his work in a minute.

现在再强调一下，我们和谷歌云的朋友合作做这个系列的原因很简单。

Now just kind of to reiterate, the reason we're doing this series in partnership with our friends at Google Cloud is simple.

因为AI转型，你经常听到很多讨论，但它仍然可能让人感觉很抽象。

Because AI transformation often, you hear a lot of talk about it, but it can still feel abstract.

比如，尽管有各种炒作，尽管每个人似乎都在谈论它，把它当作流行词，但真正展示它在真实公司、真实团队、真实约束下如何运作，并带来真实商业成果的人却少得多。

Like, despite the hype, despite everyone seeming to talk about it and use it as a buzzword, far fewer people are actually showing what this looks like in practice inside real companies with real teams, real constraints, and showing real business outcomes.

所以，我们的整个目标就是让AI转型变得更加具体。

So the entire goal here is to make AI transformation much more concrete.

我们希望你能听到其他领导者是如何应对的，他们学到了什么，他们会做出哪些不同的选择，以及你在思考自己组织内的AI应用时，可以借鉴哪些实际经验。

We want you to hear how other leaders are approaching it, what they're learning, what they would do differently, and what practical lessons you can apply as you think about AI inside your own organization.

最后再说一点。

So one final note here.

如果你是播客的常客，别担心。

If you're a regular podcast listener, don't worry.

我和保罗仍然会继续做我们常规的每周节目。

Me and Paul are still doing our thing on the regular weekly episode of the show.

保罗和凯西还在定期做他们的AI问答节目。

Paul and Kathy are still doing their periodic AI answers episodes.

所以这里没什么变化，只是你会看到更多人工智能秀的节目。

So nothing changes here except you get more episodes of the artificial intelligence show.

这要感谢这个系列，也要感谢谷歌云的赞助。

Thanks to this series and thanks to Google Cloud sponsorship.

那么，我们就开始今天的节目吧。

So with that, let's get into today's episode.

今天的节目由Gemini Enterprise赞助播出。

Today's episode is brought to you by Gemini Enterprise.

各种规模的企业都在转向AI。

Businesses of every shape and size are turning to AI.

他们正在学习如何更快地行动、做得更多，并提升自己的表现。

They're learning how to move faster, do more, and improve their performance.

但你不能随便把你的数据交给任何平台。

But you cannot just hand your data over to any platform.

你需要一个经验丰富的可信合作伙伴。

You need a trusted partner with years of experience.

这就是 Gemini Enterprise 的用武之地。

That's where Gemini Enterprise comes in.

Gemini Enterprise 能帮你构建复杂的 AI 代理，这些代理可以连接你的业务数据，同时通过世界级的安全和治理措施保护数据。

Gemini Enterprise helps you build sophisticated AI agents that can connect with your business data while also keeping it protected by world class security and governance.

所以，不再需要在性能和防护之间妥协。

So no more compromising between performance and protection.

如果你想了解更多并立即开始，请访问 cloud.google.com/gemini-enterprise。

So if you wanna learn more and get started today, go to cloud.google.com/gemini-enterprise.

网址是 cloud.google.com/gemini-enterprise。

That's cloud.google.com/gemini-enterprise.

好的。

Alright.

今天我非常兴奋。

So I'm super excited for today.

今天我们请到了约翰·科瓦尔，他是Peapack Private银行与信托公司的首席技术官。

We are talking with John Kowal, chief technology officer at Peapack Private Bank and Trust.

Peapack Private是一家精品私人银行，大约有700名员工，资产达到80亿美元。

Peapack Private is a boutique private bank with about 700 employees and $8,000,000,000 in assets.

他们为大纽约三州地区的成功人士、家庭、企业主、家族办公室及其信任的顾问提供个性化关系和定制化的银行、贷款、财富管理、投资银行和信托解决方案。

They serve successful individuals, families, business owners, family offices, and their trusted advisers through personalized relationship and bespoke banking, lending, wealth management, investment banking, and trust solutions in the greater Tristate area.

约翰今天来是因为PPAC Private从2023年开始进行AI转型，他们的故事展示了当一个组织把AI不仅仅当作聊天工具时，能取得多大的进步。

Now John is here because PPAC Private has been building an AI transformation since 2023, and their story shows just how far an organization can progress when it treats AI as far more than just a chat tool.

他们走过的这段旅程，我们今天要讨论的，始于一些AI用例，比如研究、写作、会议总结和员工辅助。

So the journey that they undertook and we're going to discuss began with AI use cases like research, writing, meeting summaries, and employee assistance.

后来扩展到了数据分析、定制应用、自动化工作流、AI代理，以及PPAC Private所称的“数字员工”。

It has since expanded into data analysis, custom applications, automated workflows, AI agents, and what PPAC Private calls digital employees.

在这个过程中，这家银行实际上在其各个业务部门建立了一个AI冠军网络，并改变了他们对软件开发方式的思考。

Now along the way, the bank actually created a network of AI champions across its business divisions and changed how it thinks about developing software.

举个例子，我们会谈到PPAC Private原本考虑购买一个供应商解决方案，成本超过30万美元，但最终他们自己开发了，因为AI改变了他们内部能够构建的东西。

In one example, we'll talk about PPAC Private considered buying a vendor solution that would have cost more than $300,000 and instead built its own because AI changed what they were able to build internally.

现在，这家银行正在构建和部署数字员工，这些数字员工结合了人工智能、传统自动化、内部数据以及面向员工的工具，来主动执行重复性工作。

Now the bank is building and deploying digital employees that combine AI, traditional automation, internal data, and employee facing tools to perform recurring work proactively.

所以，在这次对话中，我们将深入探讨所有这些内容：转型是如何演变的，PPAC Private 如何发现和开发用例，哪些技术支撑了这些工作，他们看到了哪些成果，以及 John 和他的团队在这个过程中学到了什么。

So in this conversation, we are going to unpack all of this, how the transformation evolved, how PPAC Private finds and develops use cases, what technology powers the work, what results they're seeing, and what John and his team have learned along the way.

那么，John，欢迎来到我们的节目。

So, John, welcome to the show.

我们直接进入正题吧。

Let's get right into it here.

非常感谢你邀请我。

Really appreciate you having me.

谢谢你邀请我，Mike。

Thanks for having me, Mike.

能上这个节目，我非常高兴。

Thrilled to be on the show.

在我们深入讨论之前，也许先给我们的听众简单介绍一下 PPAC Private Bank and Trust，以防他们不熟悉精品私人银行是做什么的。

So before we dive into this, maybe just give our listeners a quick picture of PPAC Private Bank and Trust in case they're not familiar with what a boutique private bank does.

是的。

Yeah.

PPAC Private成立于1921年，最初是一家新泽西州的小型社区银行。

So PPAC Private was founded back in 1921, and we started as a small New Jersey community bank.

但大约十四年前，我们的首席执行官道格·肯尼迪上任，真正地改变了这家银行。

But about fourteen years ago, Doug Kennedy, our CEO, came on board and and really transformed the bank.

所以我们从传统的社区银行转型为今天这样的精品私人银行。

So we went from that traditional community bank into the boutique private bank we are today.

对吧？

Right?

正如你提到的，我们已经发展到超过700名员工，资产达到80亿美元。

So as you mentioned, we've grown to over 700 employees, $8,000,000,000 in assets.

我们还有一个财富管理部门，管理着130亿美元的资产，并且现在已经扩展到纽约三州地区。

We also have a wealth division that manages $13,000,000,000 in assets, and we've expanded throughout the New York Tri State area now.

所以我们提供银行、贷款、财富管理和投资等服务。

So we're offering services like banking, lending, wealth management, investments.

所以我们具备大型金融机构的能力，但仍然提供非常个性化的银行服务。

So we have the capabilities of a much larger financial institution, but we still deliver a very personal banking experience.

对吧？

Right?

具体来说，我们每一位客户都有一个专属联系人，他们可以随时打电话，这个人了解他们，也了解他们的财务关系。

So specifically, every single one of our clients has a single point of contact who they can call, who knows them, who understands their financial relationship.

你知道，我觉得这就是我们所说的精品私人银行的意思。

You know, I think that's what we mean when we say, you know, we're a boutique private bank.

至于我自己，我在这家银行担任技术领导已经十一年了。

And and regarding myself, you know, I've been in technology leadership here at the bank for eleven years.

在加入银行之前，我在银行业咨询领域工作。

So prior to joining the bank, was in banking industry consulting.

所以银行业技术基本上贯穿了我的整个职业生涯。

So banking technology has been really my entire career.

我负责的是所谓的传统技术领域。

And I oversee what you'd call traditional technology.

对吧？

Right?

技术支持、基础设施、培训。

Technology support, infrastructure, training.

同时我也负责我们的技术创新团队。

And I also see I oversee our technology innovation groups.

那里有数据工程团队和AI工程团队。

That's where we have a data engineering team and our AI engineering team.

这两个团队在我十一年前刚来的时候肯定是不存在的。

And those are certainly two teams that did not exist when I started here eleven years ago.

让我们从头开始聊吧。

Let's kind of start at the beginning here.

过去几年，你们银行经历了一场非常有意思的AI转型。

So you've got a really interesting AI transformation that has happened across the bank over the last few years.

那是怎么开始的呢？

How did that get started?

是什么让它开始的？

What kicked it off?

跟我们讲讲启动之后的发展过程吧。

Kinda walk us through maybe the steps of where it went after kickoff.

对。

Yeah.

我们早在2023年就开始了，这让我们稍微领先了一步。

We we started back in 2023, which gave us a bit of an early start.

那时候微软的Copilot还叫Bing Chat for Enterprise。

So this was back when Microsoft Copilot was still called Bing Chat for Enterprise.

感觉那好像是好久以前的事了。

So it seems like seems like forever ago.

但生成式AI当时显然是新事物，我们看着它，觉得这里潜力巨大。

But generative AI was obviously new, and we looked at him sitting and said, there's just a ton of tremendous potential here.

对吧？

Right?

所以我们说，好吧，我们需要坐下来，作为整个组织来学习。

So we said, well, we need to sit down and learn as an organization.

所以我们围绕它制定了相应的政策和治理措施。

So we put the appropriate policies and governance around it.

我们还做了大量的员工培训。

We did some extensive employee training.

到2023年底，银行的每位员工都能使用AI聊天和AI会议摘要技术，而这对我们来说只是个开始。

And by the end of twenty twenty three, every employee at the bank had access to AI chat and AI meeting summary technology, and that that was sort of just the beginning for us.

从那时起，我们开始在全行建立由29位AI倡导者组成的网络。

From there, we started building that network of 29 AI champions across the bank.

这些人分布在各个业务部门，他们了解部门情况，并且正在识别真正值得解决的问题。

So these are people embedded within the different lines of business who understand the department, and they're identifying the the real problems worth solving.

这让人工智能从技术驱动转向业务驱动。

And that transition AI from being technology driven to business driven.

所以，不再是技术部门向银行推广工具和想法，而是真正做工作的人现在在构建用例管道。

So instead of technology introducing tools and ideas to the bank, the people who are actually doing the work are now building the pipeline of use cases.

也就是从那时起，业务部门开始参与进来了。

And and that's when the business lines got involved.

我觉得，从那时起，我们的进展才真正开始加速。

That's when things, I think, really began to accelerate for us.

所以就在那个时候，我们意识到，AI的未来并不是跟一个机器人聊天，然后让它去做事情。

So it was at that time that we realized that the future of AI wasn't gonna be chatting with a bot and asking it to do things.

那不是我们的工作方式。

That's not how we work.

那不是人们的工作方式。

That's not how people work.

如果我们希望AI像我们期望的那样，真正成为团队的一部分，我们就需要改变与AI互动的方式。

And if we want AI to be a true extension of our teams like we had hoped, we need to change how we interact with AI.

就这样，我们开始接受“数字员工”这个概念。

And that's how we came to embrace this concept of of digital employees.

对吧？

Right?

正如你提到的，这些AI代理是主动工作的，它们建立在其他AI代理、自动化、数据甚至我们一些更传统的工具之上。

And as you mentioned, these are AI agents that are working proactively, and they sit on top of other AI agents, automations, data, even some of our more classic tools.

它们通过常规渠道与我们沟通，比如有紧急情况时用Teams聊天，或者有常规事项想传达时用Outlook邮件。

And they communicate with us through normal channels, like Teams chat if there's something urgent or Outlook email if it has, you know, something routine and wants to communicate.

你也可以回复这些消息，进行对话。

And you can reply to those messages too and have a conversation.

这时AI代理也会参与进来。

That's where the AI agent comes in as well.

这就是我们目前的状况。

So that's really where we're at today.

我们已经开始从个人生产力工具，逐渐转向将AI作为我们团队的延伸。

We've we've started to evolve from it as this individual productivity tool more towards AI as an extension of our teams.

现在我们的目标是，在整个银行范围内推广这一做法。

And now our goal is let's scale this across the bank.

我们接下来会深入探讨每个部分，因为你刚才提到的进展非常有意思。

We're gonna dive into each of those parts here because you followed a really interesting progression.

我觉得这差不多也是整个行业的发展方向，从最初个人尝试聊天工具，到现在真正和数字员工或智能代理并肩工作。

I think it's kind of almost where the industry has gone as well from this individual experimentation with chat to actually working hand in hand with digital employees or agents.

不过我很好奇，你听起来在这些重要的AI结构性趋势上一直走得很前，无论是很早就意识到生成式AI会火，还是意识到我们从聊天转向智能代理。

But I'm curious that through line here sounds like you've been pretty early to these really important structural trends in AI, whether it's recognizing early on that generative AI was going to be a thing, whether it's recognizing that, okay, we're moving from chat to agents.

你能不能多跟我们讲讲，是什么在推动这一切？

Could you maybe just tell us a little more about what's driving that?

是你和你的团队走在了这些趋势的前面吗？

Is that you and your team being ahead of these trends?

还是领导层的功劳？

Is this leadership?

所有这些创新和保持领先，在组织里是怎么推动起来的？

How is all of that innovation and staying ahead of the curve being driven in in the organization?

我们提到了领导层，我觉得领导的支持和认可在这里非常关键。

We mentioned leadership, and I I think leadership buy in and support is is critical here.

我们的CEO道格·肯尼迪从一开始就支持这些项目。

Our CEO, Doug Kennedy, has been supportive of these initiatives from day one.

其实，AI冠军这个概念最初就是他的主意。

In fact, the AI champion concept was actually his idea.

对。

Mhmm.

所以我觉得，这就是推动采纳的关键——领导层的支持，还有愿意接受失败。

So that's really, I think, how you drive adoption with with leadership buy in and also being willing to fail.

对吧？

Right?

你肯定会失败的。

You you are gonna fail.

你会尝试一些技术，试验一些用例，但结果可能不成功。

You're gonna work with technologies and experiment with use cases that aren't gonna work out.

这很重要，但要营造这种文化很难，尤其是在我们这样的大公司里管理项目的时候。

And and that's that's important, but that's that's a hard culture to drive, especially when, you know, we're in a larger company managing projects.

项目有开始日期和结束日期，还要向董事会汇报。

And they have start dates and they have finish dates and they report up to the board.

而且，你知道，它们有指示灯，进展顺利就是绿色，有风险就是黄色，失败就是红色。

And, you know, they have little they're green if they're going well, they're yellow if they're at risk, and they're red if they're failing.

你需要融入那种文化，失败是可以接受的。

And you need to get into that culture of it's okay to fail.

尝试这些新技术是可以的，因为不然你会开始落后。

It's okay to work with these new technologies because you will start to fall behind.

你知道，我们也可以谈谈前提条件，对吧。

You know, we we could talk about prerequisites, right, too.

我们大约五年前建了数据仓库。

So we built our data warehouse about five years ago.

对。

Mhmm.

如果没有那个，我们就无法开始在那数据基础上构建一些AI工具。

And if we didn't have that, we wouldn't have been able to start building some of our AI tools on top of that data.

这些事情需要时间。

And these things take time.

所以你不可能一边建数据仓库，一边搞智能体AI，一边又在团队里培养开发能力。

So you can't build a data warehouse at the same time you're building agentic AI, at the same time you're bringing on development capability within your teams.

这些事情都需要一步一步来，慢慢积累。

These things all need to happen literally and be built up over time.

那我们就深入聊聊AI转型的各个方面吧。

So we'll dive a little deeper into each aspect of this AI transformation.

但在那之前，我很好奇，你能不能给我们举一些具体的例子，说说因为你们提到的那些举措，AI现在是怎么改变银行内部实际工作方式的？

But before we get to that, I'm just curious maybe if you could give us a snapshot today of some concrete examples of how AI is now changing the way work actually gets done inside the bank due to some of the initiatives you've mentioned.

对。

Yeah.

我们注意到一个有趣的现象，今年是我们银行使用的第三方技术数量首次比去年减少了。

So one one interesting thing we saw is that this was the first year that the number of third party technologies our bank used actually decreased from the prior year.

所以我们的AI工程团队已经开始扭转这种“买还是建”的局面，尤其是对我们这种规模的公司来说。

So our AI engineering team has has kind of started to reverse that build versus buy equation, especially for a company our size.

我觉得从历史上看，一家700人的银行，通常都是买而不是自己建。

I think historically, a 700 person bank, it was always going be buy not build.

对吧？

Right?

我们以前根本没有足够的开发资源来为每个业务需求定制解决方案，而AI开发确实正在改变这种局面。

We simply didn't have the development resources to to build those customized solutions for every business need, and AI development is is really changing that dynamic for us.

不过，别误会我的意思。

Now now don't get me wrong.

我们有很多很好的现成产品，但像我们这样的公司，如果无法构建出完全符合我们自身需求或客户需求的东西，那在当今这个技术驱动的世界里，与那些规模大得多的同行相比，就会成为一个真正的劣势。

We have a lot of great off the shelf products, But the inability for a company like us to build something that precisely fits our needs or precisely fits our client needs, that was starting to become a real disadvantage for us compared to our much larger peers in in today's kind of tech driven world.

对吧？

Right?

所以，游戏规则已经变了。

So so the game has changed.

我举个例子，就是我们在财富管理业务中的账户审查流程。

And I'd say one example is our account review process in our wealth business.

这个流程以前是用电子表格来管理的，而且它已经无法扩展了。

This is a process that was being managed in spreadsheets, and and it wasn't gonna scale.

对吧？

Right?

所以我们看了第三方解决方案，那会花我们，你知道的，正如你提到的，每年37.5万美元以上。

So so we looked at third party solutions that would have cost us, you know, as you mentioned, $375,000 plus a year.

是的。

Yeah.

而且那是一笔不小的投资，用来替换一个，说实话，在电子表格里运行得还算不错的东西。

And that's a significant investment to replace something that, you know, frankly, was working reasonably well in spreadsheets.

但快进到，你知道的，这个AI时代的黎明，我们的AI工程师用Codex构建了一个迷你平台，来组织审核和审批，并从我们的数据仓库中拉取数据。

But fast forward to the, you know, dawn of this AI era, and our AI engineers use Codex to build a mini platform that organizes the the reviews and approvals and pulls in data from our data warehouse.

再说一次，那些前提条件都在。

Again, those prerequisites are there.

而且每次审核比以前节省了50%的时间。

And yet each review is taking 50% less time than before.

同时，所以这是一个有效的软件产品。

And at the same so it's a software product that works.

同时，我们还节省了几十万美元。

At the same time, we're saving hundreds of thousands of dollars.

通过这类自动化，我们就能让员工回到客户身边，而不是整天对着电脑，对吧？

And that's how we get our employees, you know, back to their clients and away from their computers, right, by doing these kind of automations.

我再给你们讲一些关于我们数字员工的有趣细节，以及它们带来的影响。

I'll give you some interesting specifics about our our digital employees and the impact that they've had too.

我们IT部门有个叫Alex的数字员工，它负责监控我们所有的ServiceNow工单，也就是我们的支持队列。

So we've had we have a digital employee in IT named Alex that actually monitors all of our ServiceNow tickets, our support queue.

对吧？

Right?

它会逐一回复每个工单，并给出即时建议。

And it's replying to each one with their immediate suggestion.

它还会把工单转给相应的专家。

It's routing tickets to the appropriate subject matter experts.

同时，它也在持续监控这些工单队列，看是否需要升级处理。

It's also continuously monitoring those ticket queues to see if there's any escalation needed.

其实我们几周前才刚推出Alex。

So we only launched Alex actually a few weeks ago.

对。

Mhmm.

它还在学习阶段，但已经能以全职员工一半的速度处理工单了，这是个很好的开始。

So it's still learning, but it's already operating at a rate of it's closing roughly half the tickets of a full time employee, which is which is a great start.

就连我们市场部的数字员工，也会每周审查我们所有的网站。

Even our marketing teams, digital employee, does a weekly review of all of our websites.

它会查找过时的链接和过时的内容。

And it's looking for outdated links and dead, you know, outdated content.

它还会提供AEO、SEO、GEO的建议，让大家更容易找到我们。

It's also providing AEO, SEO, GEO suggestions so people can better find us.

市面上确实有产品能做这种网站扫描，但我们也只是花了几分钟，给AI代理编程，就得到了这么好的效果。

And and there are products out there that do this type of website scanning, but, you know, we also just took a few minutes, programmed the AI agent, and we're getting these phenomenal results.

你会喜欢这个的。

So you'll you'll like this.

他们的数字员工叫Mia，是营销智能代理的缩写。

Their digital employee is named Mia, which is the acronym for marketing intelligence agents.

所以很有创意。

So very creative.

营销部门的创意果然不让人意外。

No surprise creativity from our marketing department.

确实。

Sure.

但你知道，这些就是我们日常在做的事情。

But, you know, these are these are the projects that we're working day to day.

我再分享最后一个例子，规模要大得多。

I'll share with you one last example, much larger in scale.

我们和一家AI公司Verapath合作，想让我们AI工程团队的工作，借助一家更有能力的公司来增强，处理一些真正的大用例，并腾出时间。

We partnered with an AI company, Verapath, and we wanted to take the work that our AI engineering team was doing and augment it with really a a more skilled company that can tackle some really big use cases and set aside that time.

所以今天，我们实际上正在推出一个重新设计的贷款结清流程。

So today, we're actually launching a redesigned loan closing process.

你知道吗，你敢相信一笔贷款结清大概有55个不同的步骤吗？

And, you know, would you believe a loan closing is about 55 different steps?

你知道，这些流程真的很庞大。

You know, these these processes are big.

有时候你根本想不到，但我们都知道，组织里的变革管理从来都不容易。

You you sometimes you wouldn't even think it, but the change management we know in organizations is never easy.

而这就是为什么这些项目有点像多步骤的过程。

And and here's why these pros projects are kind of a multistep process.

对吧？

Right?

现在的流程是什么样的？

What's the process today?

我们希望未来的流程变成什么样？

What do we want the process to be tomorrow?

我们怎么用AI和自动化来改进那个未来的流程？

How can we use AI and automation to improve that future state process?

然后我们实际上还得实施新的流程。

And then we actually have to implement the new process.

所以这些项目很难，你知道，传统上这个行业的失败率很高。

So these are tough projects, know, the high rate of failure traditionally in the industry.

对。

Mhmm.

但是有了AI，和那些有资质的公司合作，我们就可以开始处理那些真正大的用例了。

But with AI and working with qualified companies, you know, we can start tackling those really big use cases.

所以我们可能稍后在这一集里再详细聊聊这个，但听起来这不仅仅是把技术应用到现有的工作流程上，虽然那也很重要。

So we'll talk maybe a little bit more about this later in the episode, but it sounds like it's not just trying to apply technology to existing workflows, though that's very important.

这实际上是重新构想和重新设计这些工作是如何运作的。

It's actually reimagining and reengineering how some of this work works.

对吗？

Is that right?

是的。

Yeah.

你知道吗，我们很幸运，公司正在成长。

You know, we're fortunate to be a growing company.

而且，我们现在到了这个阶段，就是有一些社区银行的流程，不太适合我们这家私人银行，也不符合我们想要成为的样子。

And, you know, we're at the point now where, you know, we have these kind of community bank processes that don't quite fit into the private bank that we are and and that we want to be.

所以我们本来就得经历这次业务流程转型。

So we had to go through this business process transformation anyway.

AI来得正是时候，我们可以说，让我们来转型这个业务流程。

AI kinda came at the right time where we can say, let's transform this business process.

同时，也把它自动化。

Also, let's automate it.

对。

Mhmm.

所以对我们来说，真是天时地利人和，现在我们可以专注于重新审视银行里的这些流程。

So it just the stars align for us pretty nicely, and and now we're able to focus on revisiting these processes throughout the bank.

这是这家银行有史以来最大的一次业务转型。

This is the biggest business transformation process that this bank has ever undertaken.

而人工智能正在推动这一切，让我们能够做一些非常了不起的事情，这些事将使我们成为一家非常有竞争力的银行，甚至已经让我们超越了那些规模是我们十倍、二十倍的同行。

And AI is fueling it and allowing us to do some really incredible things that are gonna make us and are already making us a very competitive bank even against peers that are ten, twenty x our size.

哇。

Wow.

那我们稍微深入了解一下这里使用的技术吧。

So let's open the hood a bit on the technology being used here.

你之前简单提到了Codex。

You had mentioned Codex briefly.

听起来好像有智能体在开发中，可能是在利用现有的系统。

It sounds like there's agents, obviously, in the works maybe using existing systems.

你能尽量给我们讲讲，是哪些AI模型、工具或平台在支撑这一切吗？

Can you give us a sense as much as you can share about what AI models, tools, or platforms are powering this?

比如，给我们讲讲它们是怎么整合在一起的。

Like, how give us a sense of how that fits together.

好的。

Yeah.

所以我们很早就决定主要采用OpenAI的技术。

So we we made a decision early on to standardize primarily on OpenAI's technology.

员工用ChatGPT企业版，开发团队用Codex。

So ChatGPT Enterprise for our employees, Codex for development teams.

我们建的那些智能体，很多都用他们的企业API。

We use a lot of their enterprise APIs for all those agents we've built.

但就像我说的，还有很多非AI技术，比如我们自建的企业数据仓库。

But as I mentioned, there's also a lot of non AI technologies, our enterprise data warehouse that that we built.

然后我们把AI加在上面。

Now we took AI and placed it on top of that.

这个产品叫Project Atlas，里面有我们整个数据仓库的结构，还有我们公司实际运作的知识库。

It's a it's a product we call Project Atlas, and it has our entire data warehouse schema along with the knowledge base of how our organization actually operates.

所以它能把问题转化成查询代码，然后对数据仓库运行。

So it kinda turns questions into query code that can be run against the data warehouse.

我们的数据分析师都在用，确实大大加快了获取答案和完成报告的速度。

And it's used by our data analysts, and it's really started to significantly change how quickly we can get answers and how quickly we can deliver on reporting.

你知道吗，我们现在开会的时候，就能直接回答高管们提出的问题，这真的很棒。

You know, we're sitting in meetings actually able to return answers to questions that that our executives have, which is which is a really neat thing.

而且就在那组数据分析师里，光是过去一个月，他们和Atlas的交互次数就超过了两千次。

And even amongst that small group of data analysts, there were over, I think, 2,000 transactions with Atlas between, you know, the analysts actually communicating with it in the past month alone.

所以你看，这些工具的价值已经很明显了。

So it shows, you know, how valuable these tools have become.

再跟你分享一下硬件方面，我们的数字员工其实都有自己的电脑。

I'll share with you too on the hardware side or our digital employees actually have their own computers.

你的听众可能不会觉得意外，当然是Mac mini，毕竟它在AI圈子里已经很流行了。

And your listeners probably won't be surprised that they're they're Mac minis, of course, have become very popular in the AI community.

我们可没造成这个缺货，但我们……

We didn't cause this shortage, but we

我正想说，这都是你的错。

I I was gonna say, this is your fault.

对。

Yeah.

但是，你知道，他们给我们的数字员工，在必要的时候，提供他们自己的计算环境，这样他们就能运行工具，和我们互动。

But, you know, they they give our digital employees, where necessary, their own computing environments so they can run tools and interact with us.

所以这真的让数字员工技术有了实实在在的存在感。

So it really creates a a very physical presence at digital employee technology.

太好了。

Excellent.

听起来，你五年前就开始搭建的那个数据仓库层，真的很有远见，是这一切的关键。

So it sounds like that data warehouse layer that you started kind of very presciently, you know, five years ago was really key to this.

但在此基础上，你还有OpenAI的技术在驱动和整合，听起来是和现有的工具集成，比如你用来聊天的那些东西。

But then on top of that, you've got OpenAI technology basically powering this and integrating, it sounds like, with existing tools, like things like that you're using to chat, things like that.

是像Microsoft Teams、Outlook那样的东西，还是那种生态系统？

Is that, like, Microsoft Teams, Outlook kind of that thing or that type of ecosystem?

对。

Right.

我们希望他们像我们一样工作。

We want them to work how we work.

所以他们用的是微软Teams。

So they are using Microsoft Teams.

你知道，你不需要去找聊天机器人来找到这些AI代理。

You know, you don't have to go to a chatbot to to find these AI agents.

而且记住，他们会主动联系你。

And remember, they're reaching out proactively.

所以如果有紧急情况，比如他们看了一份报告，发现了问题，他们会通过Teams发消息，而不是发一封你可能晚些时候才看到的邮件。

So if if there's something urgent, like they look at a report, they find a concern, they're gonna send a message over Teams instead of an email that you might not see till later.

你需要在那里找到一个好的平衡，但这就是我们使用这些工具的方式，就像员工一样。

And you want you want to build a fine balance there, but that's that's how we use those tools just like a like an employee would.

你说得对。

And you are right.

你知道，这些数字员工使用的技术中有很多AI，但也有一些确定性的工作流程。

You know, there's a lot of AI in what these digital employees are using, but there's also some deterministic workflows.

我们希望在AI擅长的领域使用AI，在合理的地方使用，比如总结内容、做研究、分析文档。

And we wanna use AI where AI is good, where it makes sense, you know, summarizing things, doing research, analyzing documents.

但如果我们本质上要运行相当于Python脚本的东西，数字员工可以调用它，然后直接运行那个脚本。

But if we're gonna essentially run what's the equivalent of a Python script, the digital employee can call that and and just have that run instead.

你知道，我们不需要过度使用AI，只是为了让人觉得数字员工完全是AI。

You know, we don't need to overuse AI just to just to, you know, make it seem like digital employees are fully AI.

他们使用的是适合所分配任务的工具。

They're using the tools that make sense for the task that they've been assigned.

所以明确一下，如果我在PPAC Private工作，至少在现在和未来的某些团队中，我可能会收到数字员工发来的Teams消息，主动提醒我一些可能与工作相关的事情。

So just to be clear, if I'm working at PPAC Private in certain teams at least today and in the future, I might be getting Teams messages from digital employees that are proactively alerting me of different things that might be relevant to my work.

当然。

Absolutely.

你会的。

You you would be.

你知道，这就是现在的状况。

You know, that's today.

对，是的。

And Yeah.

随着这个项目的扩展，未来这真的会覆盖到每一位员工。

As we grow this program, that really will be every employee in the future.

你知道，我们是一家规模不小的银行，每个岗位的员工都承担着多重职责。我们想让每个人回到他们擅长的领域，做他们专业背景相关的工作，把那些单调乏味的任务去掉。

It's you know, we're the size bank where everyone in the role has multiple responsibilities, And we wanna get everyone back to the things that they're skilled at, you know, the things that they have a background in and and get rid of the monotonous work.

而这正是这些数字员工发挥作用的地方。

And that's really where these digital employees are coming in.

这确实在改变我们的文化。

And it's it's really changing, you know, our culture.

我们花更多时间跟客户在一起，能够把精力重新集中在我们擅长的事情上。

We're spending more time with clients, and we're able to kind of refocus our efforts on the things that we're good at.

我很喜欢这一点。

I love that.

这引出了我的另一个问题，因为无论是数字员工，还是进行这种转型，随着AI成为银行不可或缺的一部分，你们是如何组织人员和职责，让这种转型不仅仅局限于IT或技术型人才？

And that kind of leads to another question I have because, you know, whether it's digital employees or, you know, undergoing this type of transformation, as AI becomes such an integral part of the bank, how are you organizing people and responsibilities so that this transformation spreads just beyond IT or tech savvy people?

我知道你提到了那29位AI倡导者。

I know you mentioned the 29 AI champions.

我很想多听听这个。

I'd love to hear a little more about that.

AI是怎么在整个组织里推广开来的？

How is AI being diffused across the organization?

对。

Yeah.

我们其实正在庆祝AI冠军项目一周年。

We're actually celebrating one year of our AI champions program.

他们已经在银行完成了大约100个项目，而且还有80多个项目在筹备中。

So they've they've since completed around a 100 projects at the bank, and they they have a pipeline of about 80 plus more.

这个项目管道还在不断增长。

It's a growing pipeline.

所以你可以看到，他们给我们带来了很多好点子，而且这不是管理委员会。

So you can see that they're bringing some great ideas to us, and it's not a management committee.

我们已经有够多的委员会了。

We have we have enough of those.

你知道，我们找的是那些拥抱技术、对新科技有浓厚兴趣、对AI感兴趣的人。

You know, it's it's we look for people that are embracing technology that have a a, you know, keen interest in new technology that that are interested in AI.

每个部门都有这样的人。

And every division has these people.

你只需要找到他们。

You just have to find them.

结果这变成了一个竞争激烈的委员会。

And it actually ended up becoming a competitive, you know, committee.

大家都想加入。

People wanted to join.

他们甚至很早就想把AI作为自己职业的一部分。

They wanted to make AI a part of their careers even early on.

我坚信每个组织都有这样的人。

And I'm a real firm believer that every organization has these individuals.

你只需要找到他们，他们会帮你拥抱这项技术。

You just have to find them, and and they will help you embrace the technology.

那么，迈克，你提到了管理层的支持。

And now, Mike, you mentioned management buying.

对吧？

Right?

这也是关键。

That's key here as well.

我提到过我们的CEO一直是AI冠军项目的坚定支持者，他的持续支持绝对至关重要。

I mentioned our CEO has has really been a champion of the AI Champions program, and and his ongoing support has been absolutely critical.

但你要明白，没有文化转型，就不会有AI转型。

But, you know, really, make no mistake, you won't have AI transformation without cultural transformation.

对吧？

Right?

这是一个巨大的转变，培训也是其中很重要的一部分。

This is a big shift, and training has been a big part of this as well too.

我们有所谓的科技创新网络研讨会，这是一个季度系列。

We have what we call our tech innovation webinar, and it's quarterly series.

我们在演播室和总部直播，我现在就在总部。

We run out of the studio and headquarters, which is actually where I am right now.

我们会邀请几位AI冠军参加这个网络研讨会，分享他们的成功故事。

And we bring on actually several of the AI champions onto that webinar, and they share their success stories.

所以不只是技术部门在宣传。

So it's not just technology telling people about it.

而且这总能让大家思考，某种程度上，他们听到其他部门怎么用AI，真的会受到启发。

And and it always gets people thinking, you know, in some ways, they're just genuinely inspired to hear what other departments are doing with AI.

但另一方面，他们看到某个流程，会说，嘿。

But in other ways, they they look at a process and they say, hey.

哇。

Wow.

我们部门也能用这个，这样就能传播开，让大家知道AI能做什么。

Our department can use that too, and it kinda spreads the word about what's possible.

所以总的来说，跟上技术步伐，持续学习，真的非常重要。

So I you know, overall, it's just really important to keep pace and keep learning with this technology.

而且我们每月与AI冠军会面时，实际上会与每个人单独交流，讨论他们的项目，讨论他们的工作室产品。

And when we do meet monthly with the AI champions, we actually meet individually with each one to talk about their initiatives, talk about their studio prod projects.

你知道，在我们银行里，关于AI的会议有数百场。

You know, there's there's hundreds of meetings happening about AI throughout our bank.

但如果没有这些，你就会开始落后，开始失去焦点。

But without that, you you start falling behind, you start losing focus.

这就是这项技术对我们有多重要。

That's how important this technology is to us.

为了确保我真正理解这些具体是怎么运作的。

And just to make sure I really understand kind of how the nuts and bolts of this are working.

那么，关于AI冠军，他们分布在不同的职能部门和团队中，这些人基本上是主动还是协作地把AI用例带给你和你的团队，然后由你们来构建或部署？

So with the AI champions, these are embedded in different functions and teams, people who are essentially are they proactively or collaboratively bringing AI use cases to you and your team to then build or deploy?

还是说这是怎么运作的？

Or how does that work?

对。

Right.

你知道，随着时间的推移，我们看到他们开始能够自己构建用例，比如一些简单的像聊天机器人这样的东西。

I I you know, over time, we've seen them starting to be able to build their own use cases if it's, for example, something simple like a chatbot.

你知道，我们有好几个AI冠军自己构建了这样的用例。

You know, we've had several AI champions kind of build their own.

我们很高兴看到他们向这种能力转变，就是能在自己的部门里真正创建用例。

We love to see that shift towards that ability to, you know, do actually create use cases within their departments.

但技术肯定可以成为指导，帮助他们实现自己的想法。

But technology can certainly be a guide in helping them implement their ideas.

所以，这实际上就是它的结构。

And so that's really how it's structured.

你知道，我们和他们见面，但他们带来用例。

You know, we're meeting with them, but they're bringing the use cases.

有时候我们说，你们不需要给我们答案，但把你们部门遇到的挑战带来，我们一起解决，看看怎么用AI来解决。

And sometimes we say, you know, you don't need to bring us the answer, but bring us the challenge that your department's having and let's solve it and see how we can use AI to solve it.

而且，你知道，也许不是AI，也许是其他类型的自动化，但这些对话对那些部门以及我们整个银行如何看待流程产生了很大的影响。

And and, you know, again, maybe it's not AI, maybe it's a different type of automation, but these conversations have been really impactful on those departments and how how we're looking at processes throughout our bank.

稍微拉远一点看，听起来你在AI方面显然取得了很大的成功。

So zooming out a bit, it sounds like you're clearly having tons of success with AI.

我很好奇，你是否在特定领域看到了这种转型带来的非常明显的影响。

I'm curious if you're seeing the really clear impact from this transformation in specific areas.

你是怎么衡量它是否有效的？

How are you measuring whether it's working?

你显然一开始就省下了几十万美元，那些钱本来是要买软件的。

You obviously saved hundreds of thousands of dollars out of the gate on software you would have otherwise bought.

你能不能给我们讲讲这是怎么运作的，是否有效，以及你是怎么衡量的？

Could you maybe give us a sense of how this is working, whether it's working, and how you're measuring that?

对。

Yeah.

我们刚才聊到了创造、构建和编码的能力，这对我们来说影响巨大。

So we talked a little bit about the ability to create and build and code, and and that's had a a huge impact for us.

那我们就来具体谈谈如何衡量这种影响吧。

And and let's talk a little bit about actually measuring that impact.