Your Pilot-to-Profit Playbook | Financial Services Keynote, Dreamforce 2026
In this Dreamforce 2026 Financial Services keynote, Salesforce leaders Ryan Ratting and Erin Agrios, together with partners from Anthropic, Rocket Pro, and Bank of America, outline how the industry can move AI from pilot projects to profitable, production‑grade solutions. The talk begins with a reality check on the rapid rise of consumer AI use (from 1 in 10 to 5 in 10 Americans seeking AI‑driven financial advice) and the massive $84 trillion wealth transfer to Gen Z, framing the urgency for banks to stay in the conversation. Three core questions are presented: how to appear in AI‑driven customer interactions in a trusted, regulator‑compliant way; how to untangle legacy processes without exploding budgets; and how to demonstrate real ROI. The session showcases the deterministic Financial Services Cloud as the trusted data layer, the integration of Anthropic’s Claude model via a “trust layer,” and the new headless capabilities of the platform that expose CRM data to any surface. Live demos illustrate a customer named Jennifer using Claude to consolidate accounts, a headless Slack‑based workflow at Rocket Pro, and Agentforce Operations automatically mapping, diagnosing, and rebuilding a tangled onboarding process. Bank of America’s enterprise‑wide AI strategy is highlighted, emphasizing scale, governance, and measurable productivity gains (e.g., 4 hours saved per meeting). The keynote concludes with product‑level innovations such as Coworker, Run My Day, and Meeting Concierge that simplify agent interactions, accelerate employee productivity, and generate measurable business value.
Chapters
Please welcome SVP, Amer Financial Services, Salesforce, Ryan Ratting.
All right.
All right. Good morning, everybody. Welcome to the Financial Services Keynote here at Dreamforce 2026. My name is Ryan Ratting. I lead our North American Financial Services Enterprise organization. And this is my— almost my 17th year here at Salesforce. This is my 15th Dreamforce. And ready for this? My first Dreamforce keynote. How about that? How about that? Right? So we are all in this together this morning. But actually, we've been in this together for quite some time now because throughout my career at Salesforce, I've had the privilege of working alongside our financial services customers, and I wanna thank you. I wanna thank you all for such an honor and a privilege to be able to do that. I wanna thank you for your partnership, and most importantly, I wanna thank you for trusting us all through all these years. And if you think about it, the AI conversation in financial services has been taking place really looking inward, right? How do we make our bankers, how do we make our advisors more productive? How do we lower our cost to serve? And those are all relevant, great starting points, but it's evolving, right? Something fundamentally is changing. Our customers have access to the same AI tools now, right? Think about it. We're all using Claude. We're all using Gemini, we're all using ChatGPT. We're getting more and more comfortable using them. We're getting savvier and savvier. We're trusting them more. And as a, as a result, our expectations are on the rise. And so look at how fast this is happening, right? Last year, 1 in 10 Americans turned to AI for financial advice. Fast forward to today. 5 in 10. Think about that, 400% increase since the last time all of us were in this room together. Your customers are not waiting to use AI. They're already using it. So this is all happening during a massive wealth transfer, right? $84 trillion are being transferred to millennials and Gen Z. I picked the wrong decade to be born in, that is for sure. But this $84 trillion with a T getting transferred to millennials and Gen Z, and 77% of Gen Z are already using AI to drive their financial decisions. They're hooking their AI into their customer accounts. So soon AI is going to know your customers, better than you do. That is going to present a question for all of us here in the room, right? What are you going to do in a world where your customers are turning to AI first instead of you during their financial journey? How are you going to ensure that you're still a part of the conversation? That's what we're gonna talk about here this morning. But first, how about a quick reality check? So people ask me all the time, Ryan, where are you based? And I tell 'em Delta, because I am flying, zigzagging all over North America, visiting with our customers, visiting with all of you. And quite frankly, that's the best part of the job. But I hear the same challenge over and over again, right? They say, Ryan, we have plenty of AI in pilot. How do we move this from pilot to production? How do we get real business value? And what we're seeing is that AI stuck in pilot, it is impossible for you to be able to connect with your customers where they're already going, right? AI in pilot is never gonna meet this massive opportunity, this massive moment that we all have in front of us. So how do you move forward? Well, it really comes down to how well you can answer these 3 questions, right? When customers use AI for financial decisions, how do you show up in that conversation in a trusted way? Not just for your customers, but the regulators as well. Can't forget about them, right? 2, how do you untangle your processes without blowing up your budget or exhausting your people's time? And then 3, what is the real ROI, right? How does AI truly move your business forward? So we're gonna, we're gonna help you address those questions here this morning. And the good news is we're gonna do it with proof, not with promises, right? And we're gonna hear from a great partner today, Anthropic, on the model side. And then we're very excited as well about some great friends that we have that are gonna join us from Rocket, and Bank of America. So to help us tackle this first question, how about a warm welcome for a very good friend of mine and Head of Financial Services Product, Erin Agrios.
Thank you. Good morning, everyone. Let's jump right into it. Ryan asked that question: when your customers turn to AI models for financial advice, how do you show up in the conversation in a trusted way. Here's 2 stats that should make everybody pause in the room. Right now, 57% of all web traffic isn't human. It's automated. It's AI models, it's the chatbots, it's your agents. And what does this mean for your business? This means that the customers that are knocking on the front door, your digital front door, They're not people. They're AI agents acting on behalf of your customers. And this is happening today. And we know that AI provides us with a whole bunch of intelligence, right? You heard Ryan talk about that. But in our industry, AI and intelligence without predictability is a pretty big liability. If a customer asks for financial advice, it can't be probably right. It has to be right. It has to be something that you can stand behind. And so you need a CRM that's deterministic. You need a CRM that knows you, your business, your customers. It understands the workflow that you have, the KYC processes, the onboarding processes, the claims. It needs to understand your rules the way that it should behave. And more importantly, and Ryan mentioned this, we can't forget about the regulators. They need an audit trail that they can actually trust. And that's what that CRM provides. And we've been building that determinism into the platform for the last 10 years with Financial Services Cloud. And I do want to take a moment because this year, I always tell my team, this year it's 2 children and 1 product later, we're celebrating 10 years. And that's thanks to all of you. Anybody who has submitted a product request, or joined a product advisory council, or come to one of our keynotes like all of you, thank you. Because we have built a technology, an application, a platform that is ready for us in this moment. And the innovation hasn't stopped. You heard Dario and Mark in the keynote yesterday talking about the partnership with Anthropic. And in financial services, we've been very— working very closely with them as well. And I'm very happy to welcome to the stage my friend, Greg Jacoby, to tell us a little bit more about what we've been doing together.
Greg.
Hello, hello.
Hi.
How are you?
Okay, so you're 25— I don't want to age you here, but 25-year-old— 25 years in banking, 5 years at Salesforce. You now run the banking practice at Anthropic. And I'd love for you— you've talked to many of the same folks in this room. in that role. Can you tell me a little bit about what you're hearing from our customers? Like, what are they doing with you?
Yeah, there's, there's really 3 things that they're doing. And on the theme around from pilot to profit, it's the first is maybe an obvious one. It's leadership, but it's showing itself in a non-obvious way. This is not about leadership and a mandate or sponsoring AI. We see where firms, where leaders are actually using this technology to change how they work themselves. They are, first of all, getting a clearer picture of their business. They're understanding how the technology can be leveraged, and they're able to then challenge and push their teams on how they ultimately need to change the business. And, and it helps when you see the boss being more effective. Uh, that's something where most of the people in the organization say, I, I need to figure out and how to get on board with that. The second thing is this order of operations. We found that it's people, process, products. And what I mean by that is enabling the people, giving them the tools, helping them understand how to use them, and making them smarter. When they're smarter, they can ultimately re-engineer the processes inside of the organization, which opens up the opportunity to build entirely new products and services. So this is about getting a little bit better every single day on that journey, riding the exponential. And when you do that, when you get 1% better every day, Hmm. that's 37x better by the end of the year. I think the bankers in the room understand that compounding interest is the 8th wonder of the world, and this is compounding intelligence. And then the 3rd thing that we see is Everybody here already has the superpower to scale AI responsibly. The operational risk frameworks that everybody has and has been building out time and time again, we found, is actually the accelerant to the rapid deployment of AI as opposed to the thing that holds you back. Because you can, when you can trust your audit trail, when you can trust your governance structure, when you can trust the context data you have, you can move very quickly. That's Claude running in your trust layer. That's Financial Services Cloud.
Yeah. Yeah, we talked about that, and I think that's a really good unlock. Okay, so we live like 20 minutes from each other, so we have an ongoing text chain. But lately, it's been full of the work that we've been doing together with Claude and Financial Services Cloud. Can you tell us a little bit about that?
Yeah, so true story, your team, about 2 weeks ago, got me a Financial Services Cloud org with the new Salesforce connector ready to go and plugged in. And I hooked that up to Claude Cowork. And maybe a little disclaimer here, I probably couldn't admin my way out of a paper bag. But we got those 2 things connected. And first question, you know, one shot, tell me about our favorite friend, Rachel Morris. Claude was able to discover everything in the Financial Services Cloud org, all the households, the customer objects, the contact records, all the custom objects and custom fields that had been created. And it built this incredibly rich and compelling story of Rachel and everything about the relationship that she had with our fictitious Cumulus Bank. And also sort of asked to prep for me. And I remember taking a screenshot of those things and really cool and sort of texting to you. And I don't know, I don't really know if you remember what you said to me.
No.
Nope.
What did I say to you?
Well, you said, can this actually do anything?
Oh, yeah.
Which I think was the right question, right? It's like there's all this really rich data. And so, yeah, we updated the contact records. We updated— we went and looked outside of Salesforce for all the interesting information we could find and reasoned over it and updated the CRM and made what was in Salesforce better, making the system of record the system of record. And that— I find that incredibly powerful. And then the last thing we did, which I think for the admins in the room is probably one of the more interesting things, was I then asked Claude to discover all of the processes that exist inside of the org. And it found all of the out-of-the-box service processes that your team has built. And so I picked one. I said, OK, let's put Rachel on a travel notification. She's going to take a vacation. And unfortunately, when I said, go do that, it errored out, you know, brand new org, somebody hadn't configured it. And so I asked Claude, like, go figure out what happened. It unpacked the metadata, understood what was going on, and said, you know, it was a dev org. I had admin privileges. We were just playing around. I said, do you want me to fix that? Yes, just fix it. I, that's not something that would've taken me hours to do. I, I don't know how to do that. I simply would not have been able to do that. And so I'm just incredibly excited about, you give advisors, you give bankers, you give service reps, you give admins this power of frontier intelligence with your trusted data and the things that they're gonna be able to do with that. And I was talking about 1% better every day.
Yeah.
Who knows? Like maybe we're 2% better every day with, with this, with this combination.
Yeah, that was a fun exercise. And we're gonna do more of that.
Yes.
Okay. Last question. We're, we've been talking a little bit, Ryan talked about it earlier, customers using AI, your model for financial advice, other things. I know you do first-party research. Can you tell us a little bit about like what you're seeing them do?
Sure. And the, the short answer is a lot. We, we publish something called the Economic Index. It's available online. The, the dataset is available for anyone to go explore if you want to explore this information for yourself. And the first thing we see is that when people use the consumer version of Claude, about half of their conversations are about work-related topics, about half are personal topics. The 4th most popular topic that people are talking to Claude about is financial advice. That's the 4th most popular thing that's happening in America. And what are they, what are they talking to Claude about? So first they're talking, as you might guess, investment advice.
Mm-hmm.
They're running the numbers, they're pulling their spreadsheets, but they're also doing research on a fund. You know, what, what do— is this aligned with my investment goals? Should I invest in this? And that signals to me that there's still very much this room and desire for advice. The second thing that they're doing is they're talking about the big decisions that they have. And what's interesting about that is these are conversations that ultimately lead to money, but 6 times out of 1 time, they're talking about the decision. What car should I buy? Where do I want to move? And it's an important reminder for us in the financial services industry. I think we know this, but it's a, it's a key reminder. Like, people don't wake up in the morning and say, I want a mortgage. They wake up and say, I want to give my kids a better life. And I think if I move to this town, I might be able to make that happen. And, and how can I do that? And so that's, that's incredibly powerful. And then the third thing, and this is really encouraging to me, is there's a lot of financial education in Claude. And the reason that's encouraging to me is I think probably many of you are at organizations where you've done things around financial education and maybe it hasn't quite landed. And there's something about talking with Claude where It's in the moment, the context of the decision you're making. There's the data. You can sort of run the numbers. You can ask these questions at 11:00 PM at night. You can ask it 3 different ways, and you are helping get educated. How do, what is amortization and how does that work?
Mm-hmm.
And so that's, that's an incredibly powerful thing that Claude is able to do. And, and I'll tell you the thing that I wonder and, and, Ryan, you were mentioning this earlier, is, you know, that's an amazing conversation between Claude and the customer. And I sort of wonder what that looks like when it's Claude, the customer, and their trusted relationships and advisors are sort of entering the chat and in those conversations as well.
Awesome.
Thank you. Thank you, Greg.
Thank you.
Really appreciate it.
Okay.
You heard us talk about this, and this is going to come out as a little bit of a joke as you read this slide. An AI model can't walk into a bank alone. I mean, and it's true, and I'm not going to tell you a bad dad joke, but you just heard from Greg, right? You need the power of AI and you need the power of CRM. And that's how we're going to show you in this demo here. And before I do that, I'm going to talk about the fact that Financial Services Cloud is now headless. And so you're going to see it in the demo, but we've basically exposed that decade of depth. It's no longer trapped behind a UI. So the content, the customers, the workflow, all of that now is available on any surface. So I'm actually going to show you that in Claude. And in order to do that, I'd like to welcome for the first time Ravish and Marilee, who are our demo drivers.
Hi.
Yay.
Okay. Our customer, Jennifer, she's using Claude, and she's hooked up all her financial accounts, and she wants to consolidate those accounts to one bank. So she asks Claude for some financial advice. And what Claude's doing is it's scouring the internet. It's looking for the best offer for Jennifer. And you can see here, it's going to take a second, but there's going to be 2 offers that come up. Right there, you can see Cumulus Bank. It said, yep, this is the best one. Why? It has great products, but more importantly, it has an AI agent that can communicate directly with Claude. So Jennifer doesn't have to move from this workspace. She can do everything right here. But how does that work? Well, Cumulus Bank uses Salesforce, and that allows them to offer a secure MCP connection right here. So with the toggle of that switch, Now, Cumulus can join the conversation. And so right now, you can see that Jennifer is asking, okay, great, connect. But before we do that, before we allow Salesforce and Cumulus to enter the conversation, we have to actually walk through a couple of steps because security and data privacy are top of mind. And so just like that, Claude walks her through that. And now you'll see the banking help agent. This is Cumulus's banking help agent entering the conversation. And because Banking Help Agent is connected into Salesforce, it knows the process, the needs analysis process. So, it's asking the right questions to Jennifer to really personalize that offer. So, just like that, it's got all of the information about Jennifer that it needs, and it's going back into Salesforce, and it's coming back with a very personalized offer for Jennifer based on her financial needs. So, it looks pretty good, but you can see here right at the top, Cumulus is offering her a personal banker and a personal wealth advisor that she doesn't have access to at any of the other financial institutions. So she wants to move forward. And just like that, the banking help agent can look back into Salesforce and it says, okay, I know what this referral process is. I have to offer a meeting and hook her up with the right banker. And so just like that, that meeting is booked. So what's happening behind the scenes here? Let's pull back the curtain. And what you're gonna see in Agentforce Studio is that banking help agent right at the top. That is the agent that has been working and connecting directly through Claude with that MCP connection. But behind the scenes, it's been quietly orchestrating all of those specialized agents. The needs analysis agent, the referral agent. Think about the processes that you have in Salesforce today and how you can expose them through these agents in a secure and trusted way because it understands the guardrails and the policies and procedures that are in that deterministic CRM. So that's what it looks like in Claude. You heard in Marc's keynote, though, we announced AI Force. And what does that mean? That means that you can take that experience onto any surface. But the most important point I want you to remember is it's that same platform that is governed and trusted that's gonna allow you to do that. So that's one example of headless, but I wanna talk about another example of headless. And in order to do that, please welcome to the stage Matt Baldwin from Rocket. Thanks for joining us, Matt.
Thanks, Erin.
Happy to be here.
Of course, of course.
Let's jump right in. You've been doing a lot with headless, and you've made incredible progress. And as the SVP of engineering, you're like right in it, right? But talk about it with the audience. Like, describe what you're doing. What is headless, and who's using it at Rocket?
Well, first, let me take a step back and introduce myself and talk a little bit about Rocket Pro. So I'm Matt Baldwin. I lead engineering for Rocket Pro. Rocket Pro is the B2B lending division of Rocket Mortgage. So our clients are financial institutions, independent mortgage brokers, and our mission is to help them help their clients achieve the dream of homeownership. We do that through a variety of ways. We build digital experiences, we build API integrations, we also have to build tooling for our internal team members to better serve our partners. Salesforce sits at the center of all of that. So, if you think about a team member's day-to-day today, they're tied to their desktop. They're in Lightning UX, they're hunting, pecking, trying to find data, answering phone calls, and hoping to do the best job they can to take care of their partner's needs. We decided that wasn't good enough. So, we invited Salesforce on site over the summer for a 2-day headless CRM innovation days. We had 12 ideas. We took those 12 ideas and turned them into 12 working POCs. One of those POCs was a Slack connection to our instance of Salesforce. We decided to enable it. It was quick and dirty, very easy, but we also built a few skills to go along with it, and we launched it in late August. Now, Rocket Pro has something called the Rocket Pro Experience. It's in early September. We invite a ton of partners on site, a ton of prospects, and for the first time, our AEs were able to kind of work the crowd, using just their mobile devices through Slackbot, they weren't tied to their laptops. So when they met a new prospect, they could just take a picture of their business card in Slackbot, and that created a new lead object in Salesforce. When they wanted to send out an application to that partner, they did it through Slackbot as well. They never had to open their laptop. And so we quickly saw the power of headless. You mentioned it being able to be in any surface, anywhere we wanted it to be.
Mm-hmm.
It showed its worth there. We're expanding the skills now because it's proven so beneficial. And you also think about our RAs, they're out in the field all the time. They had to lug their laptop around. Now they can do their research on the elevator up if they forgot to do it, and they're ready to go.
That's awesome. That's incredible. Yeah, I mean, it's been great. And one thing that you said resonated with me as we were kind of talking about this is like they didn't have to worry about all the different technology. They could actually just focus on the conversation they were having.
Yeah, it does reinforce, I think, a human superpower that makes human connections. And again, that's been a huge value add for us.
That's incredible. So you said your AEs, right? Your team members. What value is this bringing for them? Can you talk— you talked a little bit about it, but I'd love to talk a little bit more and just— it's changed the way that they've worked is what you've been saying.
That's right. Before I get to that, I, I do wanna take a step back and talk about probably a very familiar problem that a lot of us in this room have. We had a decade-old Salesforce instance, a ton of technical debt. And so when we wanted to layer on AI, or even when we wanted to move and make changes, it was slow. And I personally did not wanna put AI on top of a shaky foundation. So a year ago, we worked with Salesforce and came up with a plan. And rebuilt the foundation over like a several-month period. And then last spring, we started layering AI on top of it. So we implemented Agentforce, again, to allow our AEs to quickly get into our instance of Salesforce and find the answer they're looking for. They didn't have to click through a bunch of screens. We built something called ProIQ and Next Best Action. So when they were going to make a call to a client, they could be informed on like, this is what you should follow up with them on. This is the thing that happened. They were in the portal 2 hours ago. Talk to them about it.
Wow.
That sort of thing. And case management, we just started loading in call transcriptions and call summaries again, so we can capture what's important to our partners and making sure that we're able to follow up with them. Say they had a complaint about a product we didn't offer. When the product goes live, all of a sudden now we can call that partner back and be like, hey, you mentioned this in March. We're ready to go now. So it's really changing the way we work for our internal team members.
Like incredibly responsive is what I take away, right? Just based on all the data that's coming in. Incredible work. I know it's been a journey, and I really thank you for the partnership.
Absolutely. My pleasure.
Thank you.
Okay, I think we sufficiently answered question 1, right? Your AI, your clients go to models for financial advice. How do you show up in a trusted way?
Right?
We showed you a couple of examples, but one of the things I talked about in the demo was process needs analysis.
I talked about referrals.
There's a whole bunch of other ones that you have. And I know you're probably looking at me going, have you seen my processes, Erin? They are not ready for AI. And in order to help us address this next question, I'm Very happy to welcome to the stage my friend Nora Stevens Kittner.
Thank you, Erin, and thank you, Matt. That was a great conversation. As Erin mentioned, a big part of AI readiness, or the reason that you might be getting stuck, is because of all of that infrastructure underneath the use cases that Matt just mentioned or that we saw in the demo earlier. In fact, 2 in 3 financial services firms say that the reason that their AI pilot, or that their AI is stuck in pilot, is not because of the model, it's not because of the use case, but it's because of all of the scattered data and processes underneath. So that's what we're going to do right now is answer question 2. How do you untangle your processes in a way that doesn't burn through your team's budget and through your team's time. Now, back in the day, meaning just yesterday, when you'd be approaching a technology implementation project, maybe you want to improve the experience for your customers or create a solution, you would buy Salesforce licenses, you would spend hours in a room whiteboarding, you would figure out what tools are a part of the process, and you'd probably whiteboard something that looks like this. What this new era of AI has brought us is a different way of approaching this problem. Now, Salesforce has made a few acquisitions, real investments, and it's changed how we deploy technology. So for a few moments, we're going to go under the hood. We're going to take a look at how you can take these processes, gain visibility into them, and then we're going to show you technology on how to fix it. So let's go ahead and we'll hop back to Marilyn Ravish for our demo, and we're going to go back to our customer Jennifer. Now, in the last demo, we saw that she met with a banker, the meeting goes well, and she's ready to onboard. Now, we're going to show you this process fully optimized, and then we're going to work backwards and we're going to show you how we got there. So Jennifer receives a welcome email from Cumulus Bank. And she wants to stay in this, in this solution. She wants to stay in her email. She doesn't want to go to a different portal or another system. So she uploads her documents, her proof of residence, her ID, her account details. She go ahead and hit send. Now let's look behind the scenes. Let's look at what this looks like for the bank when that email lands. You'll see the full process flow in just a minute. And you're going to see a status bar change. You're going to see some colors change. And that's because we have an autonomous agent that is working in the background, and it's taking these documents and extracting data from those documents. And based on that data, it triggers different parts of the process. So you see that here. Now, if we continue scrolling, You can see Jennifer's email triggers it being passed on to research, and then to risk screening and KYC, and the process continues until Jennifer is onboarded. Pretty great, right? Pretty seamless, pretty easy. Now, for a lot of you, I'd imagine that this isn't really what your onboarding process looks like. I would imagine it actually looks like a tangled mess, maybe something like this. complex handoffs and processes, and it's really hard to get a sense of what these are. So let's work backwards. Let's go back in time and let's see how we can untangle this process map so that you can create that seamless experience that we just saw with Jennifer in that email. Now, you might be thinking to yourself, you know, how do I get this process into the system. You can whiteboard a process, you can upload a PDF of your SOPs, of your, or your statement of processes, and it gets translated into Agentforce Operations. I don't expect your operations team to take a look at this map and know what to do next. So we turn to Agentforce and we ask it, what are the bottlenecks in this process and how can we make it better? It makes sense, right? You can ask models, you can ask ChatGPT, Gemini, all of these, Claude, all of these different things. Why can't you ask it where there are bottlenecks in the process? So it identifies these bottlenecks here in orange. This is why your customers are getting stuck. This is why it's taking your customers weeks, sometimes months to onboard. And so what we do is we ask Agentforce to, the agent then puts together a series of recommendations for how this process should actually flow. And once we know those moments, now it's time to rebuild them. So the agent takes those recommendations and it puts them in— it takes those recommendations and rebuilds the process that we saw streamlined in the email. So it places in that autonomous agent that we saw that was extracting information from those attached documents. It brings in the data models, the workflows, the compliance checks, the workflows that we saw Jennifer's email moving along from research to KYC. That's placed in there automatically. For those more complex processes, like risk exceptions or approvals, that remains allocated to human involvement. We don't just allocate something to a human or auto— or excuse me, allocate something to an autonomous agent or automate something just because we can. Some things still require a human. And just like that, the process is rebuilt. Now, here's something that you don't lose once this process is live is visibility. We know that visibility is such a challenge, seeing how your processes actually run. So what this is doing is it's continuously monitoring your process And if any KPI, if your onboarding KPI or another KPI like claims starts to slip, it gets flagged. And that whole process starts again, that flywheel that we saw on the slide. The agent will analyze the process, it'll come up with recommendations, and it will rebuild it and fix it. That is the new Salesforce. So, let's take a look at what we just— let's Think for a moment on what we just saw, right? We didn't spend hours in a room whiteboarding this process. We didn't spend millions of dollars. We let AI and Agentforce Operations take care of it. And we looked at onboarding as an exa— as one example. But you can use this for claims, you can use this for servicing, you can use this for whatever is costing you the most right now. Now think about what you could do with that funding, with that investment, how you could reinvest that into other areas of your business. For our customers, we took a multi-day onboarding process or claims process, and we cut it down into minutes. That is real growth for you and your customers. Now hopefully you're starting to be able to imagine how you can gain ROI from your AI investment. And we looked at just one customer, but what does it look like when it starts to scale to thousands of customers? To help us answer that question, I'm gonna welcome back Ryan Ratting to the stage. He gets these questions asked him every day by customers.
And so I'll pass it back over to you, Ryan.
All right. Thank you so much, Nora. That was amazing. And yes, you're right, Nora. I get that, asked that question all the time. And what we're seeing is the real ROI happens when your agents start to take action, right? When they can remove all the mundane work that surrounds your employees, surrounds your customers, and remove them off their plates. When they can route approvals, when they can handle data entry, when they can surface the right data at the right time when somebody needs it. Also, the leading organizations aren't sitting deciding between AI and human connection. They're using AI to determine when best to spend human talent. Because we all know in financial services, not every moment, just like Norris said, not every moment needs to be automated, right? We have complex moments. We have emotional moments. We have high-stakes moments. And so really, the question for all of us in this room is, how quickly can AI help us move from opportunity to action to real business value? To help us answer this question, I'd like to welcome to the stage my good friend, Ganesh Krishnan.
Welcome.
Thank you, Ryan. Ganesh, it is such an honor and pleasure to share the stage with you. Thank you for joining us. Could you take a moment and maybe talk to us about your role and how you have taken Bank of America on their AI journey across the enterprise?
Thanks, Ryan. First of all, I'm glad to join all of you here for this session. So at my role at Bank of America, I'm responsible for our global banking businesses. which constitutes small business and business banking, customers with revenue of $1 to $50 million, global commercial banking, which goes from $50 million all the way up to $2 billion, and then global corporate and investment banking, which is more than $2 billion in revenue size. Outside of that, I have responsibilities for 2 horizontal functions at the bank, the enterprise CRM platform for the bank that supports all lines of businesses, and also the enterprise credit platforms for the bank. To set the context a little bit in terms of scale, because of the second part of your question, Bank of America is one of the leading financial services organizations in the world. We serve nearly 70 million consumers and small businesses, as well as corporations, governments, institutions, and individuals all over the world. So, If you think about it in terms of we've got 211,000 employees across 8 lines of businesses that operate within the United States and also 35+ countries and jurisdictions. The reason I give that context is everything we do at the bank starts with scale and size, right? Nothing at the bank can be done in a small place. It's got to scale for us. So the resiliency aspect of it is captured within what we do. Everything has to be operationally resilient in what we deploy across the bank. So now you think about the AI journey that you talked about. Our AI journey started a decade ago with Erica that we deployed. And since then, we have had 3 billion interactions with Erica since we launched Erica from that aspect of it. So the lessons and learning coming out of Erica with the NLP-based stuff we had and going into the GenAI, side of the fence is how do we deploy things at scale and make AI an enterprise capability instead of a set of disconnected projects that we do all over the place?
Right.
That's the key part of it. For that, we have 3 questions that we answer every day whenever we start anything or build anything. First of all, we ask ourselves, are we trying to solve a real customer or a business problem? or an employee need that comes into play? Second, when we responsibly deliver the capability, is a human accountable for the outcome? That's extremely crucial for a bank of our size. And then finally, whatever we deliver, can we scale it enterprise-wide globally for what's required? So those are 3 major parts that come into play. And then we have a Bank of America AI strategy. It's like, think about 3 levels of a cake, layers of a cake. Level 1 is what we call general AI productivity for our employees. So it's deployed across all of our 200,000 employees across the bank. We use it day in and day out. A million prompts generated weekly.
Right.
Can you imagine about that, what we do across the bank? Layer 2 is where we introduce functional-level capabilities or functional AI. So the enterprise CRM platform of the bank, Salesforce fits right there into the capabilities we introduced AI within that aspect of it. And then level 3 is where we introduce complex and very sophisticated capabilities that talk to our internal systems with confidential data, external sources of data, which is where the level of governance is the highest that comes into play as part of that. So beneath all of that, we have a multimodal process that comes into play so that we're not over-reliant on a single model provider. that comes as part of that. So it's all about scale for us that we bring into play.
Yeah, it's very impressive, Ganesh. And as you've scaled and embarked on this journey, what has the feedback been from your broader end-user base?
Yeah, I think about the first and foremost feedback is people tell us, do not introduce standalone tools. Introduce AI in the context of where the work happens. So taken to point in case, an example, we have 85,000 internal associates using the Salesforce platform at the bank, the CRM platform for the bank. And these are relationship managers, bankers, financial advisors. And when we introduce AI, it has to be in the context of where they do the work within the CRM platform. So introducing Agentforce in the context of where the work happens is extremely crucial, and which is what we have done. So we deployed this capability called journey of a meeting. And that's the other thing. When we deploy something, we deploy it across lines of businesses, a reusable capability. So this need of a journey of a meeting, which constitutes of a 3-step process, like every relationship manager or an advisor prepares for a meeting, plans their day, prepares for a meeting, has the meeting. You got to summarize the meeting automatically and load it into Salesforce. You got to then have an action item automatically generated for the follow-up items that comes through the process. All of that thing is what we call a journey of a meeting end-to-end. And so we have deployed that capability across 35,000 bankers and advisors at the bank, and the feedback has been very positive coming out of that. One of the things that our power users told us very recently is, this has taken the toil out of my life, and it is now allowing me to focus on having more high-touch and high-value conversations with our customers rather than worrying about preparing for all of that information, which is resulting in further opportunities, deepening of relationships, cross-sell opportunities all coming into play because of those kind of things.
So you're starting to touch on this, Inesh. 2 years into this journey for Bank of America, what is the business value that you feel that you're starting to see?
Yeah, you think about it, everything is measured with ROI for us. Everything starts with ROI. We don't just start doing stuff and launching things. Everything starts with ROI, and we follow that ROI process through the— while we are building things and after we deploy, we measure it crazily to see if it met the requirements of what we started off with. So in context of that, this Journey for Meeting that we deployed is saving up to 4 hours of valuable time for our relationship managers and advisors. And if you put into account of what we do, that's about a million meetings annually. So you can do the math of what that is. And then that's not only the productivity gains that come into play, but now you're having more meaningful conversations and which could further deepen relationships and cause more cross-sell, which talks about the revenue generation aspect of it. That comes into play. So that's a huge part of the—
What an incredible story, Ganesh. Thank you so much for joining us today. Amazing AI at scale. Thank you so much. Thank you. Well, that was incredible. And you've heard us talk about AI, but let's see how we build it, right? How do we build AI to drive opportunity to action to real business value? So to help us do that, My good friend and a member of our product team, Mr. Greg Blausey.
All right. Thank you, Ryan, and hello, Dreamforce. Are you ready? We're going to close this thing out. All right. So, Ryan threw down a little bit of a challenge for me, right? He said we need to prove real ROI. I think we've got a few ways that we can do that, and I'm going to bring that to life for you in a demo here in a second. But there's one thing that I want to acknowledge before we jump into that. And that is that we spent a lot of time, not only this morning, but leading up today, talking about headless experiences. And not everybody in the room is headless. And in fact, you're probably using our Lightning interface still. And that is A-okay, because guess what? We've got your back. So what we— what I want to answer a little bit now is talk about, well, what would happen if those agent experiences came to life in Lightning itself? So let's have a look and jump into a demo here. I'm going to start someplace that most of us are probably familiar with, a customer page. In this case, it's our customer Jennifer Blake. We're sort of over Rachel Gregg, by the way. So here's Jennifer, and we've got our trusted side panel and agent force here. And as more and more of you have been building more agents and you're coming out of pilot and going into production, we noticed something kind of interesting happening, and that was that this list, the scroll list that you see here just got longer and longer. And not only that, but it was putting a little bit of a burden on your users because your users had to know which agent to select and then what happens if they need more than one agent, right? And so it was a burden that wasn't, wasn't what's fair on them. So that's exactly why we created Coworker. So let's hop over into Coworker and look at this clean, crisp, beautiful, and very, very simple language. And I'm typing in just like I would talk to a friend. And you can see here it understands my intent. It knows exactly what I'm doing. And more importantly, it knows which agents to bring to the party to get the job done. And so that's happening behind the scenes. And here's what I really, really want you to love about this. And that is this. As you build more agents now with Coworker, your user experience doesn't get more complicated. Instead, coworker gets more capable, right? Imagine if you're a new employee and you walk in and it's day 1 on the job. This is easy to use, so we can accelerate the time or shorten the time to productivity for employees. Now, let's jump into a point. Actually, Ganesh, you made this point talking about putting AI where work happens, right? So we're doing exactly that here in just a moment, and I'm going to jump into an agentic experience that we have recently created, and that is called Run My Day. Now, Run My Day is a prebuilt experience. In this case, I'm going to be talking about it from the standpoint of a wealth advisor. And so earlier in the coworker screen, you would have seen that we were asking to follow up on Jennifer's work and that there was a referral generated. And now here it is landing on the wealth advisor's place. You can see the work is organized by compliance, service, retention, And growth. And so this referral's sitting there waiting for the advisor to hop on, and it has actually prioritized it. So Run My Day helps the advisor stay focused on the most important work and is constantly reprioritizing, shuffling based on the context that it has of the portfolio. But let's go ahead, and I want to, like, fast forward because undoubtedly there's going to be a meeting come out of this. So what we've landed on now is something we call Meeting Concierge. And meeting concierge is another part of this prebuilt agentic advisor experience that we have. There are 3 chapters, before, during, and after the meeting. Before, pretty obvious, right? It's going to help the advisor show up to the meeting the right way, prepared, understanding the context of the client, in this case, Jennifer. During the meeting, it's never going to leave your side. It's there.
It's listening.
It's taking transcripts. You can even upload a photo of notes. You can extract the follow-ups. And speaking of follow-ups, right after the meeting, there's a lot going on. And you can see that Agentforce is here ready to help with those follow-ups and commitments that the advisor made during their meeting time that they had. And one of those things that we found out was that Jennifer's having a baby, little baby Blake. And so we know that that's going to be a really, really important opportunity at some point in the future. So we're going to create a life event. Right in FSC, and we're going to put that on Jennifer's timeline, creating an opportunity for growth in the future when baby Blake arrives. We're also able to follow up with commitments that were made during the meeting. We've got some paperwork, some research on a fund, and we've even got a follow-up meeting that was scheduled. Awesome. So a lot of administrative burden that you've seen lifted off the plate. Another way that we could help drive ROI. Now let's go a little bit more meta here. And by meta, I don't mean Meta, I mean meta. And so let's click up and let's take a look at the entire portfolio. So we've been focused on Jennifer so far, but, you know, this advisor's been listening to his customers, said, man, I feel like people are saying very similar things during their meetings. And so again, very simple language. I'm asking Agentforce to go and actually summarize my last 30 meetings and look at what the, what I'm over and what I'm hearing over and over again from the clients. So I can actually use that to scale out an understanding of what's top concerns for my customer base as I'm looking to engage and grow with my portfolio. Merrilee, Ravish, thank you for a great job driving the demo. And let's bring it home. So, Ryan, I'm going to look at you here, buddy, because you challenged me with ROI and we shortened the time to productivity. We showed how we can do growth right now. We showed how we can actually plant growth for seeds in the future. We talked about portfolio and scaling. knowledge across that so that I can engage in a more effective way, build trust, and reduce retention. Is that enough ROI for you?
Absolutely.
All right, great. So let me share something with you because some of you might be sitting in the audience thinking, how can I get me some of that? Well, let me tell you what. What I showed you today, that agentic advisor experience is available today. It is in your org today. You can go in and play around with it and use it. We're not stopping there. This was an agentic advisor experience. We're building for other personas. You'll see retail banking, you'll see my favorite commercial banking, you'll see insurance, you'll see service. We're really excited to deliver that. Are you ready for more? We've heard you loud and clear that you want to work in Microsoft Teams, and I'm happy to share that in October, we will have this experience available in Microsoft Teams and soon to follow Slack. Unexpected applause. Thank you. All right. And last but not least, do not let your learning journey stop here today. We have roadmap sessions for you for banking, wealth asset management, for insurance. Have a look up here. Please go to these, engage the product managers. We love hearing your thoughts and your questions, and you'll have plenty of time to do that and opportunity to do that. And with that, I am going to say thank you for attending the Financial Services Keynote. Thank you.
