Salesforce, Inc. (CRM) Earnings Call Transcript & Summary

Good morning, and thank you for joining us for our Q2 FY '27 IR webinar on product adoption and momentum. I am Mark Murphy, EVP of Global Investor Relations, here with Valmik Desai on the far end, representing the IR team for Salesforce. We're very grateful to be joined by Bill Patterson, President and Chief Commercial Officer, sitting next to Valmik as well as Connor Marsden, President of Sales and Chief Consumption Officer. These are 2 very dynamic and impactful thought leaders of the industry. Gentlemen, first off, thank you for joining us this morning.

Good morning. Great to be here.

Our goal with this is to address investors' most common questions and topics leading both into and out of our earnings report last week. There are going to include AI monetization, AI force and other product and partnership announcements, including the one relating to Claude, infusion of AI across the platform. We're going to give you a couple of customer stories to try to demystify what is actually happening on the ground. Before we begin, I want to read you this disclaimer. Some of our comments today may contain forward-looking statements that are subject to risks, uncertainties and assumptions which could change. Should any of these risks materialize or should our assumptions prove to be incorrect, actual company results or outcomes could differ materially from these forward-looking statements. A description of these risks, uncertainties and assumptions and other factors that could affect our financial results or outcomes as included in our SEC filings, including our most recent report on Forms 10-K, 10-Q and any other SEC filings, except as required by law, we do not undertake any responsibility to update these forward-looking statements. So our plan is to start with a brief presentation and then we're going to jump into your questions. So please, at any time, when anything crosses your mind, feel free to submit your questions in the chat. And with that, I'm going to hand it over to Connor and then Bill.

Fantastic. Thanks, Mark, so much. So I'm so excited to be here today to talk about some of the momentum that we're having in the market today and really helping to turn our customers into agentic enterprises. And so we've probably shown this slide to you a lot, and I'm going to provide some context on what we see with our customers today. So first to ground us. As we've been in this agentic revolution, there's really 4 key aspects to make any agent successful. It's trust, action, agency and interface. And we have rearchitected our entire platform to be able to meet the moment in today's market. Starting with trust, trust in context and data that's delivered in a secure way with the right user permissions and of course, with 0 data retention as we start to integrate with the LLMs. Our action layer, that's our application layer that our customers have used for years and years service, sales, marketing, analytics. This is where we have our standard workflows, business processes or custom workflows and business processes to meet the unique needs of our customers. For example, Tableau today has 33 million semantic layers built in that we can now extend into our layer of agency. And you're pricing some new names on the list, Casey and Hunter and Marshall. As we've continued to extend our agentic footprint, we now have out-of-the-box use cases that our customers are leveraging to drive our agentic footprint. So for example, Casey, Casey takes advantage of the best of our service cloud application and the best of our data footprint underneath to provide great resolution for our customers with no human interaction. But when you do need to pass it over to a human, it's passed with the context of that conversation. So the human can pick up with their own agent to help solve that customer problem. And all of this is extended into our headless interfaces. Now whether that be our traditional Lightning interface or Coworker, which is our fastest-growing AI product that we've had to date or Slack or Teams, or there's great announcement that we made with Anthoropic with Claudeforce being be able to take the best of Claude to create customer interfaces that combine the best of Salesforce and the best of other applications into dashboards, analytics and purpose-built views into your data to help people get work done faster. A great example of this is a SharkNinja. SharkNinja is a great commerce customer, service customer, data cloud customer today on the platform. They launched a service agent together with our Shopper agent and saw a 6% increase in conversion. But then they wanted to build their own custom agent, an unboxing agent. And this unboxing agent when you opened up one of their espresso machines, would help walk a customer through how to set up the machine, how to configure it, how to make your first espresso, how to order any consumables for that item. And it required each layer. It required the agency layer. It required the service layer, and it required the trust layer with Data Cloud, all uniquely positioned. And they saw a 93% resolution rate for the customers who use that agent and only 7% had to be promoted to a human. So that represented a better customer experience and significant cost savings. And what's unique about our platform today is that when we look at the data lakes and the hyperscalers, they may have a data layer and they're building out an agency layer, but they don't have an application layer today. Or if you look at the frontier models, they certainly have an agency layer that they're building, but they don't have an application or a data layer and really understanding the metadata and the context and the semantics of the data in which they're working out. And the traditional application providers, they may be building an agency layer on top, but they don't have the robust data layer that we have underneath. We are uniquely positioned in that we have all 4. And all the providers wish that they had Slack today to drive AI together with your employees in the marketplace today. And through this investment, we're seeing an absolute -- our agent footprint rapidly expand. And so when I started as a Chief Consumption Officer in February to make sure our customers are getting value off their deployments. It was really about Slack bot, our [indiscernible], which is our internal agents and Agentforce which was the foundation for our external agents that we are providing into the marketplace. And H1 through both organic and inorganic innovation, we brought Piper and Marshall, our Momentum acquisition for Conversational AI for our salespeople. And then Coworker, which we launched at the tail end of H1. And that's going to rapidly expand into H2 with our pending Fin and Contentful acquisitions, our Marshall, Albert, which is going to be taking the best of Slack bot but extending it to non-Slack environments. But then Hunter, Casey, Page, Carter, just to name a few and then all powered with our headless and Claude applications. Now what's really interesting is we have 10,000-plus customers today that are uniquely using one of our AI products. And now we're seeing significant momentum as we're having more success of people adding a second, third and fourth AI solution from Salesforce. So we can continue to sell into our base and driving new customer acquisition, which roughly doubled since the beginning of February, the number of customers who are had our AI products in production. But now we can start to cross-sell into that base of customers to show more value from an agentic genetic standpoint from Salesforce. And what this really has helped us lead to is fantastic financial results. Our Agentforce and data ARR grew by 200-plus percent to $3.9 billion. Our Agentforce ARR is up 200% to $1.5 billion. And most impressively, and this is a measure of value and the work being done on the platform, we saw a 97% increase of our AWUs to $7 billion. Now we get asked the question all the time like who's using our platform? And is it old industry, new industries? None of the top 10 AI-native companies have chosen Salesforce to be their central nervous system in their brain for running their operations today. Now what's another thing that's enabled some of the success. So one of the things we talk a lot about here at Salesforce is speed to value with our customers. And we've made a lot of really smart investments to really unlock the value for our customers. So starting on the right-hand side of the screen are builders. Builders are our version of FTEs. We have roughly 600 today in the market that are all embedded with inside of our sales team working hands on board with our keyboard and we're going to be more than doubling this investment towards the end of the year. We really saw the value in out-of-the-box agents. That's the Casey, Hunter, Piper because when we have an out-of-the-box agent, together with our apps and our data layer, we're really able to drive fast time to value and get our agents up and running in 30 to 45 days. And then we wanted to unlock the entire developer community. And so we launched Agent Script, which allows us to plug in Agentforce into Claude or Codex so that they can use the development environment of their choice and drive deployment with inside of Agentforce. And that's why we're seeing the time to deploy significantly improve. And then just one other aspect I want to talk about, all these since we've identified every layer of our platform for our traditional applications, sales and service, we're able to use these coding tools, Claude and Codex to actually configure our Salesforce environment, and we're seeing a 40% improvement on how quickly we can get our customers on to our platform. One large major global retailer that we are working with needed to do a contact deployment with agents on the front end. Their internal engineering team gave them a quote that it was going to be 35 weeks to deploy a custom build of that solution. They chose Salesforce, we were able to get it live in in 6 weeks. It's phenomenal about how we can use the tools to deploy our platform, to bring agents along for the ride and really turn them into and the agentic enterprise. And this is really reflected in the new customer acquisition that we've been able to drive today. I briefly touched on SharkNinja. We have Dell that's using us for supply chain today. 19,000 individuals at Dell uses [indiscernible] complex supply chain requests, automate the processes, and it's saving roughly 30 hours per week per team. We talk about fast time to value, Live Nation. They just had a great festival bottle rock up in Napa Valley, and we were able to get an agent up in the live in 30 days where they had 37,000 guest interactions and they're going to be expanding this to all their festivals across the world. And then Wyndham. Wyndham deployed an agent in their contact center that saw a 25% increase and the average -- or decrease in the average handle time that they're able to deploy. And that required their application layer with agents and then on the foundation of data that we're able to bring to bear. So we're seeing fantastic momentum today with our platform, with our customers. We're seeing expanded usage of our apps and now we're starting to see customers go back to the well again and add the second, third and fourth agent solution from Salesforce. And this is going to drive monetization at every layer of our platform. And so now I'm going to hand it over to Bill to maybe talk about how we're trying to innovate from a pricing perspective.

Yes. Thank you, Connor. And look, I think one of the reasons that I love working at Salesforce, it's the slide that you see here. It's all the customers that we're bringing really into this agentic era truly transforming their business. But I can guarantee you, if you have the wrong pricing model and the wrong monetization model, they're not going to transform. In fact, they're going to be really confused. And so one of the things my team has been deeply focused on is really simplifying and streamlining a lot of our pricing and packaging opportunities really to give customers benefit at this time. Enterprise software has always been complicated. It's always been challenging to sort of price by feature or priced by size of an offering or how much scale do you want to put into your solutions. And so when we had this moment really about helping our customers, we went back to sort of the foundations of Salesforce and said, what we really need to do at this time is invent a new pricing structure that really aligns to the benefits that customers realize from this new technology that you just went through. And that's where, number one, the first area of focus for my organization is ultimately helping with better predictability, but really better value realization for customers. And so we've invented a whole new model called outcome-based pricing. Our help agents like Casey, like you talked about, are priced by the resolutions that they drive for our customers. And if they don't resolve the issue, you don't pay for the offering. And so this is truly putting our money where our mouth is around driving better value realization for customers to utilize our technologies. In addition, it's not always the case where you can have price agents on an outcome-based way because sometimes agents work across multiple [indiscernible] multiple domains, multiple offerings. And so what we've also done is really gone back to the drawing board around creating better value-based bundles and flex credits to flex and scale and stretch the platform as needed to really utilize and deploy the capabilities you need to make your business successful. As Conor [indiscernible] architecture is so pervasive now and it's so impressive, and it's accelerating every day, but really the flex credit manifestation, let's use all of those capabilities to use what's right for your business and just give you kind of more value at scale. And what we learned at this moment in time is customers are on different sort of modes and pacing of this journey. And so we need to ultimately be more flexible. We need to kind of offer new ways to sort of activate your business on Salesforce. And so this is sometimes where you utilize things like pay-as-you-go. So I just need a [indiscernible] enough of them consuming from the platform and put it into action. Or -- thank you, I use kind of deeper contract structures like our agentic enterprise licensing agreement, which gives you unlimited access to all the technology, utilize what you need, put it into kind of practice and then scale that across your business like never before. So these new structures, more predictability, more customer value, more flexibility are fundamentally transforming how we serve our customers in this moment because at the end of the day, if it's confusing, and if we're not serving our customers, it really doesn't matter. Let me build a little bit on -- behind the scenes about how we're kind of using this new headless 360 monetization structure. We've got a lot of questions about this from developers and a lot of our partners. But essentially, as I mentioned, everything is a compounding model here at Salesforce. And so flex credits serve at the foundation of every sort of consumption utility that we have in our offering. And so for -- if you just want to sort of get started use what we call our Salesforce Foundations platform and Flex credits, and that just sort of runs the meter for what you kind of access on the platform itself. One of the things we're really excited to do though is to offer kind of more of that predictability, more of that value for every user. And so we've created what we call this headless atom, which packages a lot of work capacity for every user make it easy. So maybe you and I use the software a little bit differently. You're probably a lot more intense on the software with how much kind of consumption you're driving. But essentially, that headless add-on monetizes just in a simple way for every user access to the platform, and it gives you enough capacity that you need to get your job done. So we take this headless out on, put a lot of more capacity. This is all the AW use, you need to be productive. And now that becomes a simple way for every organization to start accessing new technologies like CloudForce, the partnership we just announced with Anthropic. Now we're also going to take that headless out on and package it into our premium additions. So as you move up the value stack for Agentforce for sales or Agentforce for service or one of our Agentforce industry offerings, this capability kind of nets inside of that sort of productivity suite. And so this is where we start to see the magic start to occur. All of that AI centralized and organized by role, optimized for every user makes our users very productive with the amount of capacity they need and it's really that focused productivity they require to get their job done. And so that Agentforce for service or sales add-on also nets in our premium additions. So this is where this compounding monetization simple model, it just embeds into every prior addition, embeds into the top tier. This makes it really simple for our customers to plan, predict and scale and their use of Salesforce at this moment in time. And ultimately, the Agentforce One edition is the best of Salesforce for all roles. It's what includes our traditional applications like sales and service. It packages Slack. It packages [indiscernible]. And so ultimately, this is where the best of all of our company's offerings come together in that Agentforce One addition. And this is really exciting to see the growth of this offering because it means more customers are finding more value with the platform in this moment in time. And ultimately, at the end of the day, it's about transforming their businesses with the technology we serve them with.

I know, Bill, when I go and talk with our customers today, and there's a lot of concern over the cost of AI. And we've heard some of the stories on token maxing and things of that nature. We have the flexibility to provide unlimited agreements so we can take that. We can provide a predictable model on what the cost can be. If customers are using multiple solutions from Salesforce, you don't need to precisely predict what the consumption or anything is because you have flex credits that are fungible that can be plugged into your data solution if you end up using more of our data products or across your various agents. And so we can kind of meet the moment for wherever our customer is on the journey to provide them the right contracting and pricing to meet their requirement. That's where our breadth and scale really comes to provide comfort for our customers that they know they're going to only pay for what they're going to be using and whatever iteration that we decide to package it up in.

I was with the customer just recently and they said their spending is up 30%. And when they actually did the math, their productivity was up by about 3%. And so there's just a fundamental mismatch today between what is going on in the spending environment and then the value realization environment. Our mission is to change that. It's to be easier to be more simple, to be more valuable for every customer. So you're not counting how many tokens, you're counting how many leads that you processed or how many new orders that you sort of manage or how many cases you resolved on our platform.

And then just the last thing that has been eye-opening for me personally is users. What this model is providing, especially with headless is it's actually giving us access to more users internally of people who want to take advantage and consume the application. And it may not be the full Salesforce application. It may be a headless application that they're engaging with and consuming. And so it's actually expanding our TAM across the addressable user base inside of our customers.

Yes. I think serving customers is not limited to one department. And so the opportunity to really use this new platform, this new capability to reach into -- across the organization so that we all can work in service of our business and all work in service of our customers, it's ultimately where we're seeing a lot more value be created and a lot more sort of growth opportunity for our customers to realize on the basis of what this platform can do for them.

Yes. It really unlocks more -- it just really unlocks the system. So if someone in finance needs to address a service or a pricing issue. They can get access to the right screen to get their work done efficiently with AI guiding their journey.

Thank you so much, Connor, and Bill for all those comments. We'd now like to go ahead and get into the Q&A session, and we've left plenty of time for that. [Operator Instructions] While we queue up those questions, I want to begin with the first question, we see -- which says on Agentforce and Data 360 pricing, you've iterated through several models, is the pricing approach getting clearer for customers? And are you seeing their preferred way to consume it become clearer too?

Yes. First off, thank you for the question for that -- those that submitted online. First off, is the pricing getting more [indiscernible]? God, I hope so. We have actually kind of been iterating along the way around different moments of what the AI can do. And in some cases, like you've seen where Agentforce started as a platform where we're building a lot of capabilities and it was about building agents. We had a place in a platform kind of way -- and so that really made it -- so you were probably maybe too atomic for what you're monetizing in the offering. As we have these new worlds of out-of-box agents or complete agents, finished agents to serve a domain that can get to an outcome. It's a lot more easy to predict and scale those kind of price and monetization moments because the value exchange is clear. And so as the feedback continues to come in for our ecosystem, it is getting more clear. It is getting easier to deploy. It is getting easier for customers to sort of realize value. And that value exchange is ultimately kind of where, again, as Connor and I were talking, very different from sort of how many tokens do you need to be productive. I definitely think there's a much more clear approach. It's not limited to Agentforce. We've also been sort of simplifying our approach with Data Cloud and so some of those underlying technologies to serve the agents with. And so as I mentioned, we sure hope it's getting clearer. Our -- customer feedback has been quite strong. But again, how that measures and manifests is the more the customers adopt. That's where we see success.

It's natural when you have a kind of [indiscernible] explosion in all of these new technologies, the new surfaces, the new form factors, there has to be experimentation in the early innings, right?

Well, we always say the inventor of the ship also invented the ship rack. And so we really needed to make sure that we went through these iterations and tested what works, what scales. But ultimately, at the end of the day, our mission isn't just sort of put technology in the world and hope you get value from it. No, this is where we go back to our roots and Salesforce and say, you get value from the offering and get value from the agents and the value exchange between our customers and our company is clear. And so yes, I definitely appreciate the feedback. It's been a lot of iteration, but I think the signs for a customer benefit and growth of usage, as you've seen Connor, are starting to show the signs of a much more clear approach here.

Wonderful. Our second question, I should read the names as they're coming in, is coming from Luke Heinen of Madison Investments. And this is how are you or will you monetize Slack bot? If customers prefer to use Claude within Slack, is there a way to monetize that or does that limit the Slack revenue opportunity?

Yes. Great question. Thank you, Luke, for the question. We are planning to monetize Slack bot as part of your Slack subscription. First off, Slack is a per seat or per user subscription access fee. So everyone who sort of wants to access Slack and pays a per user per month offering there. Inside of that per user per month offering, we give you a lot of capacity to use Slack bot to be productive. So if you're asking questions about your customers, you're asking questions about sort of your suppliers, you can use Slack bot for all kinds of sort of productivity experiences. And every user gets a lot of questions that they can ask. Now if you have power users and you go beyond that sort of inherent supply, you can always add more capacity for Slack bot to do more work for you. So again, this is where we load you with enough work to get started. You utilize what we call our Flex credits to expand that usage capacity as you sort of need -- usage grows over time. And it really allows our organization to sort of kind of crawl and walk and truly run with AI into their offering. The second part of your question about embedding Anthropic inside of Slack, does that limit Slacks revenue appeal? Absolutely not. Anthropic is good at some things, Slack bot is good at some things. Other partner offerings and other AI offerings are good at some things. And so what you really should expect from Salesforce and ultimately, with Slack as a surface and the canvas for us is expanding the number of opportunities that other productivity systems can integrate because what we know is integrating in the tools of choice of the customers that we serve is going to ultimately make them more productive on the platform. So we don't see that as a limiter. We actually see it as an amplifier. And we do think that it also drives happier users. By the way, happier users drive more usage, more usage drives more retention. So this flywheel of opportunity and growth truly starts to get groundswell and [indiscernible].

So I actually had this question happened with the customer where they were using Claude and they're evaluating Slack bot. And fundamentally, Slack bot uses Claude as the underpinning of the LLM. But we provide practically unlimited capability -- unlimited use of Claude with inside the Slack bot licensing model today. And so what I told the customers, they said, hey, you can use Claude and pay per token or you can have an unlimited use of the application inside of Slack bot and not worry about your token expense, have the same capabilities, have the same ability to share skills to be productive and drive it, and they found that to be a much more appealing model than using Claude from a token perspective today, and they opted for Slack bot. And so this is, once again, from a pricing packaging standpoint, what are the scenarios where you're going to uncap the amount of use of the application and spread it across a lot of users, we're going to power users and light users and have the right model versus a scenario where you're going to be paying per token that potentially can expand where you're not getting the productivity and the customer opted for Slack bot. And that's a pattern that I see playing out over and over again.

Absolutely. If I could add 2 more points on the Slack piece because it's so important to the core strategy. The Slack interface is expanding as we speak, the innovation that the team is driving there. We've talked about Slack code briefly at earnings. When you come to Dreamforce, you're going to see a full version of what Slack code looks like for a developer. That interface that UI, where you're able to get your work done where previously you might have been switching between 4 or 5 different screens just to get that workflow complete, Slack is now becoming that hub for a lot of that work. Two other stats that I think were really impressive from the quarter. First, Slack had its strongest net new AOV performance since we closed the acquisition in Q2. And second, since we launched Slack bot, just in the period from GA to today, we've seen a tripling in upgrades. So when we think about the value of Slack and how we're thinking about this broader interface layer, the more we integrate with more partners bringing those amazing multiplayer agent experiences into that interface. The more we drive more Slack bot usage consumption adoption, the better for Slack's overall business performance and frankly, the better for how we are actually interacting with Salesforce, right? For me, I do a lot of customer meetings, when I get asked to, I'd love to do it. Previously, someone from Connor's team would come over and be like, okay, let me give you the prep. Let me pull stuff from our Salesforce org. Let me give you the full view. I don't have to do that anymore. I just ask Slack bot what should I know about x customer? What's their recent products? Are they having any customer success issues? What's the team trying to sell? And it gives me a full prep dock directly within my Slack channel. And within that prep dock, I have Connor's team coming in and telling me, hey, here's where we really want you to focus. So this flow of the interface being everything that you would have normally done across multiple different layers showing up in Slack, we think is super, super exciting.

It's a game changer. It's addictive. I can tell you that as a new joiner that never had access to Slack and had never seen it. I said recently, if you take it away from me, you're going to have to pry it out of my cold dead hands. That is how addictive that product is. Thank you for that. Next question is going to be from Alex Zukin of Wolfe. How should we think about how much certain agentic workflows cost? Is there a dollar value per lead or is there a CPQ agent or a marketing agent? How many AWUs will that burn? And how should customers think about fitting this into budgets, particularly with something like Slack and [indiscernible] channels, et cetera? Thank you for the 3 parts question there, Alex.

Alex, thank you for the thoroughness of the question. Look, let me tell you how our team thinks of setting pricing in this moment because as you've seen and heard earlier, we went through a lot of iterations on this topic. How we think about how much certain agentic workflow costs? Is it really about setting a value and a value for how much -- how valuable is the offering that we're sort of putting into the hands of our customers. For example, we know there is a dollar per lead because that's ultimately how organizations size and scale their own performance and own planning and budgeting activities for how many marketing qualified leads they need to really sort of enlist and fuel their sales funnels, if you will. So we do look at this sort of in terms of what organizations today pay, not just in software but total costs, labor costs, workflow costs, software costs, the integration costs. All of that sort of equates to some degree of utilitary budget that an organization was sort of planned for. What we're doing is we're competing on the basis that we think we can make that easier for organizations to do based when you use our technologies. And so I don't actually want you to be in -- the second part of your question about AWUs that burn, I actually don't want you to think about that at all. I think that's an internal measure of utilization for how we are sort of seeing how much performance happens out of our platform called the Agentic Work Unit. But what I want you to be thinking about and what I want your customers and the industry to be thinking about is really kind of the labor and the economic offset of it used to cost me maybe $5 to do a phone call, well, now I can actually do it for $1, $2 per resolution here on the Salesforce platform. So this is where we want to just think about kind of offsetting kind of expenses for our customers and making it much more economical, much more easy to scale using the benefit of our platform that can now serve that unit of work.

Next question is from Kirk Materne of Evercore. Can you talk about how the horizontal pricing strategy for AI complements your industry Claude pricing in any industries where these new AI pricing configurations are landing best?

Yes. This is fully a pricing webinar today, I think. The horizontal pricing strategy and the industry pricing strategy sort of follows that same structure of compounding value like we talked about where the horizontal platforms are priced sort of as a baseline and the industries and the vertical offerings that we have, where they have much more acute value and much more precise value, therefore, higher value, if you will, are priced at a premium offering for our customers. And so in terms of sort of that compounding value in that stair-stepping model, our Financial Services Cloud is priced as an uplift on top of our core sales cloud, if you will. And so this is where you see sort of that value compounding sort of structure really come to fruition. Are there industries where AI pricing [indiscernible] the best? I can give you two and then maybe, Connor, you can add in. I think financial services is an area specifically where the data sets, the regulated sort of environments, the sort of out-of-box functionality we've built for kind of KYC and household management. These are specific areas where we have price premium capabilities for sort of the financial services industry that we price on top of our core sales cloud but which maybe doesn't need those areas in the horizontal selling kind of way. Same thing on health. And again, where you see the sort of regulated environment where you have PII information, you have information that around HIPAA compliance. Our offerings that have these sort of compliant boundaries have all of this nested value on top of the horizontal offerings and so this is where you see sort of AI for industries really command that premium price point, if you will. Maybe you can add a few from your standpoint?

I mean we're seeing a lot of momentum in our consumer goods market right now with our CG cloud. And we're seeing all the leading CG players migrate over to our platform. And there's really 2 scenarios. There's a customer who already had CG Cloud today and now they're agentifying their processes with our out-of-the-box agents for CG Cloud, and we're getting an uplift as they're starting to deploy those agents. And we're seeing a lot of net new customers that are buying, deploying and creating global models for CG cloud, but they're starting in -- the starting position, they're agentifying their work processes. And with CG Cloud, the out-of-box usage -- out-of-the-box use case they're able to deploy it in 30, 45 days on top of their existing platform. And so we're seeing tremendous momentum for that right now because the AI is purpose-built for trade promotion management for retail execution. It's an agent that sits alongside the salesperson to understand what's the package that I sold to the customer, have we sent the rebate check back to them for the display that they put inside of their store. All the approvals and workflows are flow through the back end of the system that the agent drives it. And it brings in the best of both worlds, the determinism and the probalism on top of the industry solution. So we are seeing these industry models and these industry purpose-built agents really driving value and speed for our customers which is so critical as things get more competitive in the marketplace.

Wonderful. Our next question is going to be from Samik Chatterjee of JPMorgan. What are you seeing in the early days for adoption of Agentforce One edition? And do you envision the -- how do you envision the customer sales motion to upsell working?

We are seeing an explosion of upgrades to the Agentforce One addition. It's been absolutely a phenomenal motion for us. And we think it's going to accelerate because we're adding headless into the model as well. And so before, it was about unlimited agent force for the internal users. So an agent sitting alongside a contact center worker to help walk through a customer issue and to guide them through resolution, leveraging the full knowledge of the organization. And now it's going to be not only that agent sitting on top of, but then also what are the headless interfaces that, that employee is going to potentially want to leverage to drive. And so Agentforce One edition has been a big winner for us and we see it accelerating as we move through the foreseeable future.