Name: The Future of Salesforce Document Generation & Agentforce in 2025
Uploaded: 2025-01-28T17:30:02.560Z
Duration: 57 min 44 s
Description: The Future of Salesforce Document Generation & Agentforce in 2025

Transcript for "The Future of Salesforce Document Generation & Agentforce in 2025": Hello, everyone, and welcome to another Salesforce Ben webinar, this time in collaboration with S Docs. And today, we're gonna be speaking about document generation and agent force. I'll give it a few seconds just to let people dial in. We have a huge amount of people registering today, so, just letting some people flow in. Hey, Anand. How's it going? Good. How have you been? Yeah. Very well. Thank you for being here with us today. Cool. Let's get started. Anand, if you wouldn't mind sharing the presentation and, yeah, let's go. So today, we're gonna be speaking about the future of Salesforce document generation and agent force in 2025. So those that know me know document generation is probably one of my favorite, categories of ISVs in the ecosystem. I used to work in the space many, many years ago and saw the power it can have across so many different use cases, industries, practically any industry. You know, most departments who are using Salesforce can benefit, with document generation in in some way. And we are officially entering a new era with AI and agents and automation. So I'm really excited to share the stage with Anand today, to hear a bit more about how S Docs are approaching this this next, evolution of technology. So my name is Ben McCarthy. I'm CEO and founder of Salesforce Ben. Been working in the Salesforce ecosystem for 13 years this year, which is pretty crazy to say. And I'm gonna be your host for today. I'm gonna hand over to Anand in a second who's gonna take us through the presentation and demo as well. And we're gonna be having a q and a at the end. You've got your q and a panel in, on the gold car screen. So feel free to use that throughout the session, and we'll get around to as many, questions as we can at the end. So, Anand, over to you. Awesome. Thanks, Ben. Thanks for the introduction. Good morning, good evening, good afternoon, good night maybe, for those that are joining us globally. Big fan of SF Ben and Ben especially. My name's Anand. I'm the CTO for S Docs. Just wrapped up my 2nd year at S Docs. But prior to that, like Ben, I've spent, a lot of time in the Salesforce ecosystem. I spent 15 years as Salesforce's, architect in their professional services team, certified technical architect myself. So I understand the platform very, very deeply. Just kind of the reason, I came to S Docs to kind of lend all that expertise to S Docs and the Salesforce ecosystem. So so we're gonna talk a little bit about the now and the next, you know, AI and, you know, we'll talk a little bit about that as well, especially since this past weekend. I think there's a lot of, like, deep seeking that a lot of companies are doing, no pun intended. So let's talk about what is document generation. Right? All of us for those that are familiar with Salesforce are, for that matter, any kind of, like, SaaS application. It's place where you have a lot of data. There's business rules, workflows, processes, all of that. At the end of the day, any business process has these moments where you have these documents. In a typical b to b sales, it's quotes, invoices, contracts, master services agreements, NDAs. In a public sector, it's, you know, constituent letters. In health care, it's enrollment letters. So there are these places where data becomes documents, and that's in a nutshell what document generation is. It's taking data, making them documents, and adding that context. Right? And it all starts with the template. You you fill out and plug the data in. You have logical blocks in there. You personalize it. So if you're in the country of Germany, add the Germany data council clause into this contract. If you're in California at CCPA, if you're an AARP over the age of 60, then add this to your enrollment better. So it is moments of personalization. And then, obviously, this is one record at a time or many records at a time with automation. So that's, like, in a nutshell what document generation is. And as we talk about the future, let's also look at the journey that we've taken. Right? So 2,005, for those who remember, now I had not been in the ecosystem then, but 2,005 is when Salesforce officially launched AppExchange. The first few, solutions start to emerge. 2,007, 8, and 9 is kind of a big moment for document generation. That's why the first few document generation solutions start to emerge in, AppExchange. Salesforce launched the platform. It used to be called force.com, became, something else. And then, obviously, if you're familiar with Salesforce, the name has changed, but the technology gets better and better every time. And we've kind of been in this journey since 2009. I I joined in 2008, and I've kind of rode along with Apex, Aura, Lightning, Visualforce, you name it. Right? Process flows, automation. And the platform has evolved, and what it has done is also evolved this domain of document generation along with it. Right? So we we kind of been have been in the business, give or take, about a dozen years now, and we started when Aura or Lightning didn't exist. And you had Apex and Visualforce, and we've kind of grown with the platform. And, you know, throughout 20 tens, we've adopted that, and we've increased more and more of the platform usage and what we use as a platform. And 2024 and 2025 is bit kind of like this watershed moment for SaaS and cloud in general with the emergence of AI. OpenAI had its big moment in December of 20 23. And right this weekend, we're, like, going through this interesting yet another wave, for those that are watching social media about deep sea can you know, how AI is getting disrupted, not once a few years, but probably every every other week or every week or every day maybe. Right? So there is this rapid pace of innovation. COVID accelerated digital, to a large level. Right? So that's template the journey that we've been in. And, you know, based on our experience, I kinda, like, illustrated some of the use cases here, right, across sales, service, legal, finance, HR, operations, health care, manufacturing. You know, you have these different varieties. But why does all of that matter? And and, you know, I was reading through the state of Salesforce yesterday and the metric and and you can pull this up, on your own. But if you look just Google state of sales, a report by Salesforce from the 2024, 70% of a salesperson's time is still is still spent on non selling activities. 70%. Just do the math across a company that has 100 sales reps, 500 sales reps, 5,000 sales reps, and 10,000 sales reps. It's staggering the amount of money and time that is spent on not selling activities. And that's why this all matters. Right? It's productivity. Right? You meet your customer. If there is a request for a receipt or request for an enrollment letter, why have a human have to do that? Bring in automation, bring in AI. Right? And and make it personal. Drive the compliance aspect of this. Right? Because if you are in public sector or state and local, there are some certain things that you need to do and make them automated, have those moments of documentation injected into all of those. I'm always ashamed to say this, Anand. But we we've just, we've just implemented DocGen into our business because our our salespeople used to used to kinda use a master template and then kind of cut out sections that they didn't need, copy and paste in bits that they did. And they did that for every statement of work. And, obviously, it's extremely manual, very prone to error. They had to do that with the t's and c's as well. And now we've just implemented Docgem where, obviously, what products are selected in Salesforce merged in automatically. But, also, you reminded me of something really cool is the the kind of logic, the if statements you can do that only add in certain t's and c's if the products are in there. And even though we've only got 2 salespeople so that, you know, the the time saved we're not talking millions of hours here, but but still it's significant. And they can spend that extra time doing, you know, earning earning their commission, essentially. Exactly. Exactly. Right? And and and what I wanted to kinda touch upon, is the road ahead. Right? Like, you know, like, again, you you can see the the statistic that's coming from the state of sales state of work that Slack published. But, you know, it's all about productivity and efficiency. AI and automation are kind of like the enablers. They're kind of like the vehicles to which all of that efficiency is driven. Salesforce is very famous for clicks not code, but I think Salesforce is now in this moment where it's not clicks not code. It's prompts not code. We wanna kind of take take on that same queue and say, your document generation shouldn't be a click. It should just be a prompt away. Right? Or it should be an extension of a prompt that's already part of a business process. So again, prompts not good, not clicks not good. Right? Or and clicks not good. Top talent, from a skill set standpoint, you don't want to hire a sales guy that's doing 70% of the work, kinda chasing things or finishing documents or, you know, typing up updates on our document or a proposal. Right? You want them to go sell. Like, that's how you're making them motivated. They're earning their commissions and whatnot. And, you know, compliance, security, trust, trust, which is a very important value for us at S Docs. Having been at Salesforce, trust was always the number one value at Salesforce, both for Salesforce as well as ecosystem of partners and customers and, you know, consultants. And, you know, how do you kind of drive all of that data, data security? All of that is super important in especially in this day and age of AI where data is all over the place. Right? Like, where where do you you don't even know where your data is going. And and lastly, not but not the least, customers' expectations. Right? Like, customers in this day and age are at a point where if I don't have to interact and talk to someone to go get something done, I'd rather do that. Right? So we are at this crossroads where customers and users, starting with consumers, are getting comfortable with, interacting with agents, kind of these autonomous agents, and they become more human like. You know, it'll evolve. I'm not saying we take the human out of the picture, but their expectations are evolving. Their expectations of how fast these things can respond to their asks is increasing. Right? Humans, by nature, tend to be slower than systems that are much more fast. And, Anand, just just on that previous slide, I think kind of initially when you kind of look at these things, you don't really think the the solution to to your problems are documents. But why why are why is document automation so critical to solving some of these issues in 2025? Yeah. So so I'll I'll I'll start with current state. Right? Like, I mean, we're not gonna get rid of documents. I mean, at the end of the day, anything and everything is a document. At the end of it, it's the artifact that comes as a result of all the work that you're doing in sales or service and marketing or even in the industry world. Right? Like so it it is the artifact. And I think, at s stops, we're actually redefining what the document is. And I'll kind of, like, talk about, this from a few slides here. But if you think about what is going into a document, right, it's word processing. It's a bunch of text clauses, legal stuff, whatnot. There's data pulled from your CRM, pulled from Salesforce, pulled from spreadsheets, whatnot. There's logic. And then there is security. Right? Like, hey. You know, the legal team is saying you cannot change the termination clause. Or the finance team is saying don't change the revenue recognition clause, and then only revenue should be doing. And then there is the natural collaboration that's happening within the company to get a document done and across companies when you have 2 legal teams or 2, sales teams or 2 procurement teams talking to each other over a document. Right? So there's that all of that. But if you think about today's document, and I'm referring to a Google doc or a Word document or a PDF document or a Office 365 document, all of these five things are still on the person or the individuals that are most involved in that document. And and and back in the day, and I think this happens in every company of any such, size, when you're drafting out a contract, there's usually about one person, the sales guy that's driving it, but then you have legal and everything. And you as a sales guy is trying to bring everyone together, and there's, like, a 1000000 versions internally. And then you send it to the customer, and then there's, like, a 1000000 more versions. And there's this constant, like, work that you're doing, the admin overhead, to essentially keep up with the latest. And then your data is changing. Right? So you added a new product because the customer is like, hey. I saw agent force. Can we throw an agent force into the into the code? Oh, I made some updates to the code. Can you make sure that those updates stay the same? So there's this constant copy pasting fact like, there's a lot of errors that are happening. And and with this is why it's important because, again, automation and efficiency comes when you let systems manage some of this. And and today's documents are are are like, today's documents are the form of Google Docs and Word. Amazing word processors and technology, but they missed some connections to a few things. Right? So the way we're redefining the document is we think about a document, which is like the words and, numbers as a connection of layers. Right? So you've got data layer. You've got logic. You've got approvals. You've got security collaboration. And with the advent of AI, AI kind of also becomes a layer. And what I mean by this is think of for the end user, it's a simple document. It looks and feels like a Google Doc or an Office 3 65 document, but it's aware of your data. It's aware of your security. It's aware of your, approvals. So as your data changes, you should be able to simply come and say, hey. Refresh my data. And all your edits remain, but your data changes. This I mean, we we actually have this working and you know, just a slight detour even though this is a Salesforce webinar. Like, we launched our S Docs for HubSpot last year, and we actually have this working. And when we show this to our customers, and they're like, oh, what do you mean I can actually make changes and still have all my data intact? Yeah. Right? And next evolution to this, and we've heard Salesforce talk a little bit about this, about AI being the UI. We think document is the UI in the future. When when when you do a contract, you're usually referring to a contract document. And someone's like, can you send me the contract document? You're not saying, can you send me a link to the contract record that has all these fields? You're usually looking at the document, and oftentimes, you're like, control f, search for, like, you know, payment date or end date or termination class. And we believe the evolution is the document becomes yet another UI for data. It becomes an entry point for approvals. When, you know, the salesperson says, I'm gonna give them 10% discount, they're just typing that in the document, and that that triggers a approval process that the manager perhaps approves in Slack. Right? So we think that that evolution of document being the UI because, like, just just half an hour ago, I was out with a customer, and we showed them the the the product. And they're like, what do you mean you can actually see this and preview this before I send out the document with my data? And they're like, yeah. This is this is completely shifting what is the definition of a document. Does that does that help answer your question, Ben? Yeah. 100%. That's great. I mean I mean, this this sounds like quite a a big shift in terms of how you're foreseeing kind of users to interact with documents. So, I mean, that was kind of one example of of how a salesperson might interact with a document. But do you have any other other examples of how different teams could interact with, you know, this type of new way of of looking at documents? Right. And I think, you know, let let me kinda talk a little bit about the architecture of, SDAP so that kinda that that helps with this. Right? So so we are built, natively to the platform. And for those that are not familiar with the word native, what that means is that we actually don't have any server infrastructure that is outside the Salesforce platform. So we run fully within the Salesforce platform, which means that we use only Salesforce technology, Apex, Visualforce, Lightning Web Components, custom objects, and and and the likes. And we're built on top of that Salesforce data model, the Salesforce trust layer. And we have these different entry points that, you can generate. So if you're a Salesforce user, yes, you can certainly use the document as your UI. But if you're familiar with opportunities or contracts or accounts, you can just simply use our out of the box components and build it. Or say you have a very unique business process where you're collecting some information, then you want to generate a document and send for signature, you could do that using flows and our SDK. And our SDK is kind of like that common layer where you can generate a document and put your process on top of it. Right? So we have built this flexible way of interacting and generating documents as well as sending for signatures. We we have some out of the box ways of interacting with it. But if you have a bespoke and say you have a custom lightning component, that is exposed in an experience cloud page that you want a student to fill in their details, get an enrollment form, and then sign that enrollment form right after, you could do that with S Docs. Still all with your data, your logic living inside Salesforce and never leaving the Salesforce trust platform. Does it happen? Yep. That's great. Thank you, Anand. Awesome. So, again, this market texture, again, kind of the the key is we sit on top of the Salesforce platform. Right? Like, we can pull any data from anywhere in the Salesforce platform. As long as it's exposed in Apex, we can pull it. Now we can also pull data from outside of Salesforce, but the mechanism would be through the, Salesforce layer of Apex or otherwise. So as long as you could do it in Apex, you can pull that data. And we have lots of customers, that have, you know, external systems that they pull data in run time and pull it in. Right? So let's talk so we've talked about current state. We've talked about the pain points. Redefining the document. But, you know, this this webinar is about AgentForce and AI. And as I said, we're in 2025. And if you've not heard of AgentForce thus far, it's time to kind of go go into the LinkedIn because, you know, our the fearless leader, Mark Benioff, has been talking about AgentForce. It it feels like it's only been a few months, and they're already at AgentForce 2 dot o. I I fully expect that's in Trailhead DX, they'll have Agent Force 4.0. I think they almost skipped skipped an upper this month. But it's it's it's it's exploding, and there's a lot of buzz around it. And, I was in New York, over the winter, and they had a workshop for ISVs because the Salesforce recognizes the power of agent force when combined with ISVs. And we're looking at this new era of, you know, agents and autonomous things. And I think, I wanna kinda highlight 2 profound quotes that was like, wow. I mean, like, this is so deep. 1 is from, the CEO of NVIDIA, talking about how IT departments are historically, like, IT departments are the ones that are building these agents, but they kind of the agents are kind of like humans doing work, and they kind of become the HR department. So it's such a profound point of view. I think he said this at CES earlier this month. And then, of course, Mark is talking about digital labor and, you know, all the things that can become efficient, super impactful as as a company for an enterprise. I mean, Salesforce is obviously doing this. They've got agentforce and help dot salesforce.com. And I think it's a it's a profound shift in how we think about technology. A key part of this is also data. Salesforce especially is poised really well, and, you know, this comes from my experience at Salesforce. One of the things that large enterprises and, actually, companies of any size, small or big, do is that they have all of these data, that sits in custom applications or data warehouses. But Salesforce has that platform to pull that in. Now either through Apex and other things or through modern product capabilities like data cloud, they're pulling all this data. And for AI to be useful and meaningful, you have to have data. It's almost like garbage in, garbage out. If you don't have the data, no amount of, you know, even, AGI, was not gonna help you. Because if the data is not great, not gonna get greater results either. Right? And I think that's why Salesforce's agent forces in this profound like, I almost feel like in is gonna have a meaningful impact on the agent world because they already have enterprises that are pulling their data into Salesforce for their sales teams, and now they can tap into the AI, the lang large language models. And that can kind of, like, dramatically shift the productivity of the end user. So so AI, you know, like, I get a few stats here to talk about. Everyone is talking about AI. CIOs, sales teams, developer teams, everyone is starting to experiment with it. And I I almost feel like 20 2024, late 2023, early 2024 was a lot about AI experimentation. People are, like, still kicking the tires. I'm like, not really sure, but I wanna, like, try to figure out what what what this can do for me. And I think 2024 started to see those early wins, and experimentation continues because the large language models are continuing. OpenAI has launched, like, probably 4 different models. We've got other competing models like DeepSeq or, Entropix. I forget their model's name, but cloud. There's all of these models that are emerging. People are still kicking the tires with it, but there's a lot of results that are coming out of it. There's domain based AI companies that are coming. Harvey, as an example, is a legal AI. That's building specialized language models. I we're we've been talking to a a partner, a company that does, like, legal AI solutions like Harvey. You've got manufacturing. I think I just saw earlier today, LinkedIn, Reid Hoffman, who's the former CEO of LinkedIn, started the company for pharmaceutical AI. So you've got all of these profound AI, based companies that are coming up, but the results are starting to slowly emerge. I think the ROI on this whole AI thing is still kinda early days. I almost draw an analogy to this and, like, the early 2000s when cloud was starting to make its, you know, early days emergence into the technology industry. People are like, there's not a lot of ROI in cloud, and here we are 2 decades later, and if you don't have an application in cloud, people almost stared, like, which which generation are you from? Right? So so let's talk about agent 4s. Again, if you if you've not been on LinkedIn or social media, let's give you a quick preview, and and, Ben, please add to something that I might have missed. Right? Like, what is AgentForce? Like, autonomous AI agents using your own language models. Salesforce has the ability to tap into chat GPT, or you can bring your own model. And these agents are wired up to Salesforce. They're wired up to marketing. They're wired up to service cloud. They're wired up to Slack. They're wired up to Tableau. So they're bringing all of that, AI large language model power right into Salesforce. And it it caters to a couple of use cases. Right? 1st and foremost, natural language, based so prompts, not clicks. 2nd, it can do work or it can do generative AI, which is more like summarize this email or, hey. There's about 150 cons on this case. Can you just summarize the e, the case and tell me the next steps? Or what's the customer sentiment about this case? So that's the second part, which is the natural language reasoning as well as generative model. There's a part of it that Salesforce talks, which is the Einstein trust layer, where it's grounding your prompts. Because Salesforce has this trust being their number one value, they're making sure that data that you have isn't being used to train the model. And whatever personal data that you are sharing, a a Social Security number or a credit card number, are responsibly passed so that the AI is not kind of, like, doing something, you know, nefarious with it. And then because Salesforce always builds platforms, not products, your investment in the, platform is gonna, leapfrog ahead with AI because all the, programmatic customizations you've done with flows is gonna suddenly become AI enabled with AgentForce. All the programmatic, customizations you've done with Apex can now start to become AI enabled with AgentForce. So AgentForce not only comes with out of the box actions, but you can extend and build your own actions with Apex, with flows. And I'll I'll I'll get to a demo in a second. And, of course, all of this means productivity. Ben, any what are your thoughts on the course? Yeah. Great summary. Yeah. As you said, I think you need to be living under a rock. Not to have heard about Agent Force at this point. Salesforce have done a great job with with marketing this and hyping it up. But, I I think we are now starting to see I'm starting to hear more and more implementations that are just about to go live. Sales also are working really closely with some of their biggest customers to get this live, so we should see a lot of really cool use cases very soon. I've got a couple of things to add. I think, one, everyone probably remembers Einstein GPT that was released at TVX back in 2023. So I think what a lot of people forget is that a lot of stuff that was released during that period, such as prompt builder, is still extremely relevant. So prompt builder integrates with agent 4. So, you know, if the result of you chatting within let's say you've got a customer support agent and you want it to, I don't know, at the end of the conversation, summarize everything that it's done. You would actually use prompt builders to do that. You know, you give this prompt. You put in some merge fields of, you know, the the case summary and stuff like that, a case description, case subjects, and it generates something for you. So anyone that worked on those kind of tools before and started experimenting with that is still very much relevant. Agent Force doesn't replace that. It's in addition to, prompt builder and a lot of lot of those tools. The second thing I'd I'd mention is, if anyone's paying attention to the the stock market or the, tech companies at the moment like NVIDIA, like Anand alludes to at the start, DeepSeek has just been, just been released, which is a much, much cheaper version of Chat gbt's sorry. OpenAI's Chat gbt. And if you look at the stock market, every well, today, it's kind of up again, but everything yesterday was read across the board. Down, down, down. But Salesforce is actually up. And the reason for that is because people believe this is this is gonna massively benefit Salesforce. So if the AI models and the LLMs that Salesforce are using are much cheaper to run, then Salesforce should be able to offer it more cheaper. They might not. They might just make more money, but it's good for the stock price. But you'd like to think they pass it to customers as well. So, you know, that that kind of turmoil at the moment is very interesting, but people believe it should benefit Salesforce. Yeah. And I think it's specific to what happened yesterday and over the last few days since DeepSeek. Right? Like, and there's a lot of buzz in LinkedIn about this. Like, how did Salesforce pop about almost 4%. Mhmm. I think it comes down to data, and metadata. Right? Like, I think this is super important. I think Mark also tweeted about this on x. Yes. You know, Salesforce, because of the couple of decades that they've had, with enterprises, is either the system of the source of truth or the system of truth for data, and in many cases, becomes a system of layer, system of engagement, for data that's otherwise locked inside of a data warehouse or a custom application. And because Salesforce has done really well with that data layer, not just, in its, previous state, but even with data cloud with all the data virtualizations at Snowflake and Databricks and Google's BigQuery and and whatnot, Salesforce has invested in talent. And and if you were, you know, if you've been in the ecosystem last year, you would not have a Salesforce event without data cloud being uttered a 1000 times. I think data clouds, they've built that foundation, and they've built that bedrock because data is the foundation for AI to being useful. And to me, that is why I think Salesforce in this is again in this unique position, which is why commodity models like LLMs and, you know, how much cost it it to kind of, like, do reasoning engines will kind of follow. And, you know, Moore's Law, for those who who follow that back in the day when, you know, personal computers were a big deal, Moore's Law was predicting, you know, the cost of semiconductor is kind of going down every few years. And I think that's gonna happen with LLM and GPUs and with, you know, how much it is to serve an AI. And, you know, if I were to kind of, like, predict BOLI a decade from now, I think the cost of getting an AI model is gonna be so cheap that it's gonna be irrelevant. But the cost of acquiring data, making sense out of the data, unifying the data, stitching that data together is still gonna be like I mean, I I started my career back in, like, the in 19 late nineties. And data data unification quality and bringing all that data together was always a challenge. We had the Informatica, the Ab Initios, the, Dell Boobies. We had many of those tools trying to say, yeah. We're the ETL. We are the ELT, and we we still have a lot of those companies. And all of them are great companies. No digging against them. But good data quality and unified data is still an extremely large, complex, and somewhat unsatisfied, problem today in the enterprise, especially b to c enterprises where you're dealing with millions of these data points. Right? Like, you're a target or if you're like a a a Geico or, like, if you're a, you know, Bloomingdale's. You have all of these data points, and how are you bringing that unified data? Salesforce is taking that big effort with data cloud, and and there are obviously other solutions out there in the market. But I think Salesforce, because they're putting agent force, data cloud, platform extensibility, all of these layered on top, and the security model, the sharing model, all of that, it just has all of those ingredients over the last. And then the metadata model, how how you can kinda, like, describe your flow, describe your Apex, describe your custom fields, describe your objects. That, I think, is that unique value proposition for agent force. Yep. Alright. Definitely. We agree. So so how does SPAX fit into agent force? Right? So I'm I will bring this slide back, and we talked about the market texture. And to us, the opportunity was agent force becomes this extension to all the work that we've done on top, and it kind of sits as this layer on top. Right? It's yet another point of engagement for our, templates, for our, document generation capability, for our SDK, for the flows that use our invocable actions. Right? So while we still support all those entry points as we're gonna launch and embrace agent force even more, agent force becomes, hey. If you're comfortable typing a prompt budget or you get talk to makers, requesting signature, by all means, go do that. You wanna click a button? Do that as well. Right? Like, I think 2025, 26, 26 is gonna be in a a a time and maybe sooner than 2026 will be where the UI will actually start to evolve, where more and more people are gonna have simpler UIs, but it'll just be a text box like OpenAI or many of the AI tools. And you just type what you wanna say or even click the mic button and just say what you wanna say, right, wanna do. So I think that'll evolve, and agents are like, the entry point to agents is a prompt. But at the same time, there is this you know, I was talking to a friend yesterday, and there is this advanced mode of people who are still wanting to go look at the data, click a few things, ensure data integrity from a data capture. So to us, agent force becomes another entry point that is matched language enabled. Salesforce does a lot of that. But then we, almost 18 months ago almost all this coming, maybe, if I should say that. We invested in this SDK layer, which essentially exposes all of what S Docs can do as an Apex SDK so that you can call that from anywhere, whether that's another Apex class or an Apex customization or, like, in the component or from a flow or from your own screen flows or from an Apex trigger or from an Apex API. Right? Like so because we have that unique Apex SDK, we were able to embed base agent force, and, you know, I'll I'll I'll get to the demo in a second. But, you know, here's a quick gist of what this can do. But when we, when I saw agent force and, I I saw this workshop, I'm like, okay. I get it. I was able to go from idea to actually having a working prototype in under 3 days. And and part of the reason is that we had that SDK already available. Had we not had that, I would have had to go expose all that SDK layer. And I have a lot of ISV friends, and and they're kind of, like, doing this rapid thing of exposing it. And, you know, we're now, like, SDK first. And because we have that SDK, agent force was super easy for us to do. So do you wanna see a demo of this, Ben? Let's do it. Alright. Let me make sure that I'm doing this right. I'm gonna share my screen, and maybe I'll just do entire screen. So so for those that are not familiar with SSTOPs, I'm gonna give a quick tour of SSTOPs, and then I'll jump into, like, an actual demo of agent 4. So this is like a simple demo art. Again, if you think about doc gen, you have templates, data put together because of document, and then you have signatures inside of it that goes around. Right? So I've got a couple of really simple, templates. I've got an NDA. So if you're a salesperson, you kind of know it. We've got a really simple business focused template editor, if my connection would be any faster. But, again, I have a really complex NDA that has this. Right? Now in a traditional way, I'm gonna actually show this in in in how you would do this Otherwise, we have the, desktop's, out of the box lightning components, and I'm gonna do that here. Hey. I've got generated documents lightning component in here, and I've got the, the generated documents, lightning components in here, and then I can choose my templates. Let's just say I wanna do the NDA and quote. NDA and Coke. Right? Again, this this this is purely to demonstrate the native nature of our, of our our solution, and I'll get to, agent force in a second. Alright. Let's do this. I've saved it. So if you're a regular sales user, yeah, you can come select code, generate. I'm not gonna do that today. But, yeah, with agent force, we've got the agent force always active here. So I can come in here and say, generate an NDA for this deal. Right? And this is where the power comes like, yes. I can certainly go click on it, but you could kind of, like, do a natural prompt. And agent force is now, like, looking at my prompt, looking at the document, the NDA. So it detected that, hey. NDA is actually an ad stocks template. So it go fetch the template, detected that this deal it it knew that this deal referred to this edge emergency generator, opportunity. And then it brought this together, invoked the s doc SDK, and then generated the back of it. And you you can do this from anywhere, but let's just say I wanna do this in my home page. And let's let's just go to the home page, and I wanna do this. I'm gonna clear the history so that's not confused. I'm gonna say generate, a code for the edge installation deal. Right? So if I if I were a salesperson and I knew the deal, like, generate a a code for this deal, I don't have to click, go to my opportunity, click the button. It's kinda, like, saving me 6 or 7 clicks, and it's right there. Right? It's generated. It's ready. And this is just the beginning of what we're trying to do with S Docs and Agent 4. So as I said, that that SDK layer offers the ability to generate a document, email a document, send it for signature, request status on a particular, signature, and and so on and so forth. So we're, like, starting to we we built we built the basic case of, like, generating documents, but we're taking every one of our SDKs and starting to implement how do I make that an agent force prompt or prompt enabled so that you can then configure your prompt and associate it? And and then, by the way, you can do this for any object. I just showed this for opportunity, but you can do this for accounts or custom objects or contracts or invoices. So the possibilities are infinite. How does that look, Ben? That's great. I'm looking at the future right now. Thanks, Ben. So, I think that was at the end of our, at at least, flight. So maybe we can start looking at questions, and Yes. Go from there. Yep. Sounds great. Lots and lots of questions coming in. I just thought I'd clarify a couple of things. Just I saw some questions coming through. The tool that I mentioned earlier was was called deep seek. And someone had some questions about chat gbt being cheaper than agent force. That is true. So the the point was with DeepSeek is the DeepSeek is is a much cheaper tool to use than than OpenAI. So if you need to a so Salesforce or API into an LLM for the back end of Agent Force, and DeepSeek is meant to be about, I don't know, 50 times cheaper or something like that. And then I can't remember the exact stat, but Yeah. It was like yeah. Yeah. I think they they they're officially saying 5,000,000 is what they used to train, and Agent Force is somewhere in the vicinity. I sorry. OpenAI is somewhere in the vicinity of, like, 500,000,000. So order of magnitude, about a 100 times cheaper. But, again, we're just learning all this. I'm not gonna read too much into the marketing or the hype of this. I think this will all settle down and we'll kinda notice. Yeah. I'm gonna look at some of these questions. Yep. Take it away. Zach or Saima, should I just look at that the the answer and approve? But I'm gonna just take these questions and then tackle them top one top top to bottom. Alright. It's just questions coming at such a rapid pace. I might need to go bottom up. Because I think the look the most recent question comes. So is there a way to pull data which differs based on start date? I think, Tasneem is the person that asked this question. I think that's more logic. So Tasneem, the short answer is yes. I probably need to get some specifics. So, certainly, email us and, we can get to it. How will AI affect admins? Ben, do you wanna take that one? How does AI affect Salesforce admins? I mean, massively. I I think, you can approach it from a few different angles. I think, one, it's gonna affect how you work with Salesforce. I mean, if you look at, OpenAI, they've just released operator, which allows you to control a web browser. In theory, you could use that to do Salesforce implementations, you know, declarative ones, create flows, create fields. I haven't tried it myself yet, but, it's pretty scary. You've then got admins implementing AI for users as well. You know, you're gonna be implementing agent force potentially for your for your users. And then you've got, you know, you've got developer tools, AI, coding tools as well. So, I mean, it's gonna it's gonna change a lot. I think the adoption that we're seeing of certain AI tools now, especially developers using AI coding tools, is pretty crazy. So, yeah, definitely keep up to date with everything that's going on. Yeah. So I'm gonna answer a question which has, I think, come up a couple of times, first, and then we'll get to the next question. But, I think there's a question about forms, and form development, with S Docs. So today, we don't do forms. We have historically when we do planning for, the year, it comes like, is this the year that we build a forms capability? So one, I I'll answer this in 2 parts. 1, I think Salesforce flow has become so good, at forms with respect to Salesforce. Screen flows reactivate, flows. There's just so much that you can do with drag and drop that we feel like just building forms would be a bit of a reinvention of the wheel there. And thus far, we've had a lot of customers that adopt that format. Now if you really want advanced forms, we do have a partnership with, a a a few firms. One, being Eureka, who's another native ISV from Salesforce. And we do have a partnership with Eureka, and then we can kind of build that. I think that, so that that was kind of, like, the forms question, that I think, Tim, you asked. But template designer, you're clarifying with template designer. I'm gonna say, we are doing something really big about the template editor, designer. I am not at liberty to share that right now because it's an active project. But, Tim, I'm gonna follow-up with you and get connected with my product teams because we actually are reimagining the template designer, and it's you're gonna love it, but I want you to kinda give us feedback. There was a question about replacing DocuSign's CLM and signature capability. By no means am I a DocuSign expert. They're, of course, the leader in the esign space. But if you go back to the slide that I talked about reimagining the document, I think a lot of CLMs come at this from, a systems and data and, you know, process perspective. We're flipping this on its head and saying, well, your your contact is your document. You live in the document more often than you live in the system. So we're not trying to become a CLM, but at the same time, I think we're trying to become a better integrated, connected document platform that has a lot of capabilities that otherwise what a CLM would provide. So I'll I'll leave it at that. Maxwell, thanks for the question. Probably need to follow-up on that. S Docs, I think it was a question about package. We are a managed package. Go to the AppExchange, search for S Docs, and you can just install it on your sandbox to your production. There's a question about, can I add a link to the newly created code inside AgentForce? Yes. Absolutely. You can do that. In fact, in our next version of the demo, it not only generates it, but it'll actually say, here's a link to your document. You click it, download it, and email it, and we may add additional functions to do that. This question about mass generation of, documents. Yes. We can. Because we're we've got that Apex layer, of SDK. You can include that in an Apex batch. We have customers that generate millions and millions of documents in bulk. We have a customer that generates tax documents and year end financial statements, and they generate approximately I wanna say, like, 200,000 documents in a 24 hour period, towards the end of the year. So they have a lot that they do. View and modify the body of the document or the attached code. This question. This is actually one of the power features of S Docs and something that we're actually trying to incorporate through agent force. But a lot of times people wanna send a document. It's a PDF and you can't edit a PDF unless you have Adobe. But with S Docs, you can actually generate a document, edit it, make tweaks to it, and then send it. And the the version is maintained. So and we're trying to, like, what's the best way to kinda bring that forward into agent force? But with our SDK and our lightning web components, you can already do that. Reader text, using s stops and allow the user, to use text. Marco, I might need to circle back on this. Yeah. So use What does agent look like for complex doc creation? So I think this question from Stacy. Stacy, I might need to follow-up on this. I'm not sure exactly I ask. Understand your question about calling that doc for calling sorry. What does the agent look like for calling that document generation? The agent itself is, so if you understand agent force, there's topics and actions. The action itself is literally 6 lines, invoking the SDK, which we will package as part of a managed package that, you can install so that that action's available. So you don't even have to write that six lines of code. But today, if you have an agent force enabled instance, you can use the SDK to actually do this. Stacy, I'll circle back with you because we have a managed package extension that offers some basic actions out of the box, and we'll be kind of adding more and more to that. We're active in security review, and, hopefully, in in a few days, we'll actually have that approved. So there's a question of can, can agent 4 say, can you please open up that document that just generated from the home page? I believe, Jared, the answer is yes. Just need to kinda identify the best way to kinda, like, figure out the prompts and the instructions for that prompts. The question about large templates, how do we pull the right template? Budget, I think, it it all comes down to the prompt and how you design the prompts and the actions. So you can kinda, like, do specific searches or you can kinda say, hey. I found 4 templates based on what you just said. Pick the template that you're choosing. I think agent force has that capability, so it's all about the prompt design, the topic design, and the action design. That's how you think about it. And and we're looking at that to be as well. Can voice be used as a prompt? I believe you can use the standard browser based, dictation capability. I don't know if agent force has voice. As as far as I know, I don't think so. I think they're single model, text only from what I know, but they might be adding multimodal later. Priyanka has the question about is this feature available? We are in security review, so I'll have a better answer with this, Priyanka, in a couple of days, but we're hoping to have that security review be done in the next few days. And, if you're interested, reach out. Information will be shared, about this after the webinar. I'm happy to pass on my email if you guys wanna reach out directly. Can Agent 4s do the signing? Absolutely. Yes. So we started with generation for version 1. So I think the question was, can it also send, the document for signature? So v 1, we're having document generation done. And then the next evolution of this is to be able to to not only generate but also send documents for signature. Zach and Sam, how are we doing on time? I think we've got about 10 minutes or so. There's a lot of questions, so I'm, like, trying to do as much as I can. Alright. Can Agent Force do the signing functionality? We talked about that. Will email templates be generated? You certainly can. We support over IT formats. PDF, docx are kind of like the dominant formats, but you can actually generate formatted emails also through, s docs. And, depending on how you wanna manifest that, you certainly can, do that. So, Van, I think, would be useful to kind of understand your use case. For instance yeah. So there's a question from Travis, which is more about how do you mix approvals and quotes. It it's kinda like we can kinda take care of, like, the document generation part. I think for the approvals, it and, again, Agent Force is capable of doing it, where you have to kind of, like, define the prompts. Or if someone says, can you approve this document? It kind of looks at the right quote and then automatically flips the status. I know Agent Force can do it. It might be worthwhile exploring it to see what's the right prompt and action for it. Okay. So there's a question about expanding our way to sign contact sign contacts, I assume. Destiny was the question. Using personal ID cards. Yes. We do have the, we do have something and we're gonna be showcasing, some of this at the DC World War in a few weeks. We are actively working on, so a lot of the federal and state and local company, customers have this requirement of, CAC and PIF cards, which are essentially smart cards that you just tap and it verifies their identity and then completely, completes the signature. So we we're starting with that. I don't know, Destiny, if that's what you're referring to, but, there are more advanced, things like personal ID cards. I'll have to get the specifics to understand that. But we are embarking on a a project to kind of add better, support for identity confirmation. Okay. There's some questions that are related to prompts. I think the one thing with agent 4 is based on, my experience building it, that I have seen is prompt building as it as much as it feels like it's natural language. There is a little bit of, like, trial and error and, almost like art to that prompt building. This is why I think this, prompt engineering has become this growing domain that people are starting to learn about it. Because you think, like, natural language, I speak English or whatever not now native language that you may have. Why do I need to engineer a prompt? They make it sound a little more complicated. But there is, like, there's a very common term called RAG, which is reinforced augmented generative AI. And what that is is essentially in your prompt, cluing in some additional data that you already know about it and then enhancing the prompt so that the the AI and the large language model can actually respond. And that does require a level of understanding of how prompts work. Salesforce has a variety of knowledge. And I think Salesforce, Ben, you you guys, I think, did a newsletter over the weekend about, Agent Force and Promise, if I'm not mistaken. Right? I think it just I I I I saw my thoughts on Saturday. Yeah. We're we're, we're putting out more content than ever on Agent Force and everything through Salesforce. So we've got a newsletter, which you can sign up to on our on our website, and that pretty much will keep you up to date with most things in the ecosystem. Yeah. This is a great question. Can AgentForce do the templating for us at the moment? I think for simple templates, I would say yes. But I think as you get into the more complex stuff, oh, I need this image here and I want this table there or that's where this gets a little more complicated. But, again, I think with the right prompts and the right type of metadata, I don't see why not. But I wanna say that that's not a use case that we're, like, directly aiming for right away, partly because template is right once and then use it a 1000000 times. So your time saving on agent force doing a template is probably not as significant as compared to everything that comes from generating a document. Alright. I think we're almost at the end of all questions. And and in case we missed any questions, feel free to reach out. You'll get some emails for follow ups. I'm gonna, like, be bold and and put my put my email directly in here. I'm gonna go off my script. So so I'm gonna give my S Docs email. For those that, want any information about S Docs, feel free to email me. I may not have the answer right away, but I'm certainly gonna point you in the right direction. I think that's There was a Yeah. Couple more questions coming in if we have time and end in q and a. Yeah. I think, Scott, thanks for calling out docx. We are taking a very concerted effort with docx. For those that don't know, docx is, the Microsoft Word format, and, a lot of our customers do wanna generate docx. So we are, we do support docx today, and we're taking a step back and relooking at it. Salesforce actually added a new capability in their, winter release or I I forget what release we're on, but I think it's either winter or spring. And I think we're gonna, like we are relooking at that. So, Scott, I'm gonna actually circle back with you. In fact, I might connect you with my product teams to see if you have some complex stuff. And, you know, we are relooking at DOCX, so we're gonna get better and better, DOCX support. Another way to call the SDK methods. Yeah. I mean, Pano, I think that's certainly a possibility. Again, we're not looking at template creation as a immediate use case for, AgentForce as of today. But, you know, our customers and folks on this call and on this webinar are probably gonna change our our direction. So certainly let us know if there's some great use cases. Alright. I think that's Right. Almost, a wrap at this point. Fantastic. So yeah. I mean, thanks for everyone joining us. Thanks, Ben, for hosting. Any any parting words? No. Nothing for me. No. It's been really great. Thank you, Anand. Super interesting session, kind of giving us a glimpse into the future of of Agent Force. And, I mean, there's loads of use cases that we see from Salesforce, but I think this one's really unique to see how how agent force can work with ISVs. And it was a great use case, and DocGen is still one of my favorite app categories. So, yeah, it's been a pleasure. Alright. Have a great evening, Ben, and have a great evening. Thanks, everyone. Yeah. Bye.