Name: How to build AI agents that create business-ready documents
Uploaded: 2026-02-19T16:01:07.840Z
Duration: 45 min 8 s
Description: How to build AI agents that create business-ready documents

Transcript for "How to build AI agents that create business-ready documents": Hello everyone and welcome to TEMPLIFY TALKS. I'm your host as always, John Cinnaakos, Senior Principal Customer Success Manager at Templify. As you'll notice, we're in a new look studio. We're always in a new look studio. We'll keep you guessing. For those of you that haven't been on Templify Talks before, you can view all of the previous episodes on our website and somebody in the chat is going to put a link in there. Now, I mentioned the chat. This webinar is powered by Goldcast. Now, if you want to interact with us, you have a chat on the right hand side. So, we can test that now. We're coming to you from Copenhagen. So, just to test the chat, can you just type in where you're coming from? And, thank you. It's a pleasure to have you here. So today, we have, a catchy name Enable AI Agents to Create Business Ready Documents. Just rolls off the tongue. And, to have this session with me, I'm joined by two legendary guests. The first guest is a Tempify Talks noob. Now, I've been trying to get him on Tempify Talks for years and he's finally here. He's been responsible for creating some of the best automation solutions we have with our customers. He is our field CTO, Stefan Lars. Hello, sir. Welcome to Tempify Talks. Thanks. And secondly, returning to Tempify Talks, this man has been solving document creation issues since I was in nappies. He is a Project Director of Product Management and our Managing Director, Ian. Welcome back! Thank you. Very excited to be here, John. Good. Last time you were here, it was for session, so it's good that we're back together on this timely issue. So, to frame where we are, enterprises are rapidly adopting AI platforms to build and automate workflows. They're using Copilot and other ecosystems. Now, the ambition is clear. We want faster document creation, smarter workflows, and obviously scalable productivity. But therein lies the issue. When it comes to generating documents, the situation is a lot more complicated. And the reason is, AI outputs are often inconsistent and can be non compliant. Agents lack understanding. There's no way for them to understand the template that should be used or understand layouts. For example, it can't understand the brand. So, they don't inherently know which template to use, which disclaimers are required or even which specific content should be used for which regions, etc. They lack context. And the result is the opposite of what they tried to achieve. It's rework, inconsistency, and even worse, it could lead to non compliance. So today, we're going to explore how you can use AI agents to create business ready documents. The core insight is simple, but very powerful. AI cannot solve this in isolation. So, we're going to discuss the guardrails, the business rules, and how structured intelligence can go on top to give you an accurate output, no matter the document you're trying to choose. That's where TEMPLIFY comes in, and that's where we're going to talk to our guests and show you some live examples of how this can be done. During the session, there will be a poll or two. At the end of the session, you're going to get a send out, which we call the Templify Use Case Library, which you can see some of what we've spoken about today, and it will contain contact information for you to get in touch with us and learn more. So, with that in mind, Stephane, you've been a field CTO for a short amount of time as it's a new title but you've been working in solutioning for a very, very long time. With the advent of AI, how has the conversations you're having and your role changed? Thanks. What is interesting about AI is that everyone is, of course, using that to optimize workflows. And that means that we see AI as something that triggers the creation of a document workflow more often than we've seen before. In the past, processes were maybe a little bit more structured, And AI allows for more flexibility and more interpretation of things you as an organisation might receive. Think about an RFP, for instance, that needs to be analysed. And AI can do that, but it also means that that needs to be picked up again by something and go through a process. And we call that the AI enterprise gap. We're good, yeah. Okay, perfect. So, what are some of the challenges, like obviously the solutions we used to have? We used to have the data, a template and an output. Obviously, that's much more difficult to do when you're introducing AI. So, what are some of the conversations you're having about what solutions are trying to be created and what problems are trying to be solved? What we see is that enterprises, who are a little bit ahead of the rest, are adopting solutions like Copart Studio, but also Clean and Blockbringing, for instance, where it's about enterprise search, it's about bringing data together and pulling that from all kinds of sources. But that usually leads to a chat. Yeah. And not necessarily to a document. So, it is then our task to take in that context and turn it into a business document that can be taken to the next step of a process. And when you say the next step of a process, obviously, there still going to be, in this example, collaboration. So, you'd need to use this sort of office tools, documents. It can't just be a static text. Well, what I mean is usually if you produce a proposal that is sent for a potential customer. So you receive an RFP, you need to analyze that RFP, figure out how you answer that RFP with your proposal, and we then help turn that RFP into a proposal that makes sense for the services or products that you offer as an organization. And that document can then be shared with your prospective customer. Awesome. And, Ian, in your example, or what you're going to show in a moment, is that extra layer on top of obviously the AI can create and pull data, but it's that the part that really sells the proposal is the branding. It's that mark of quality that the recognizable sections from the brand's perspective that will essentially help close the business. The better a document looks, very often, as long as it's aligned with the brand, it can help the performance. So, I guess, why is that part so important compared to just the data within it? Well, to be unique against your competitors, you want to be precise on what you have as output. So the documents that we create and that could be either an onboarding deck or a proposal or a pitch, it should be on brand as you decided within your company from your brand perspective or marketing perspectives, And what we can fill here the gap indeed is that this data that's collected from everywhere, from the user, from the chat, from a data source like CRM or Dynamics in Microsoft, we can bring it together into that performing document. And that's not possible by AI so far as we can see, because it's not able to create a high fidelity PowerPoint deck or Word document with that high quality that you want to have repetitively, because AI creates maybe good stuff, but it's not repeating the same thing. And as a company, want to have this similar thing that's always the same if you go out for a specific customer to win your deal. Okay. And obviously, we can there are certain, as you mentioned before, we discussed around your solutioning. There are certain things that we can do, including AI, that doesn't necessarily need the power of Copilot or Gemini or those kind of things. But that extra layer is obviously a lot of these organizations, as we've talked about at the start, have started to adopt these tools and that gets them so far. So, what would you like to show us that can sort of frame this conversation a little better? So, before we can jump into the super advanced stuff, we first need to do a little bit of education on, or maybe a reminder of what Templafy is. So bear with me for just one minute where I talk you through our compilation engine, because that's in the end needed to be able to generate any document, whether that's from an AI triggered workflow or more rules based workflow. So let me show what Templafy currently looks like. What you see here is PowerPoint. And what I do is I find one of my templates. In this case, I go for an audit proposal. And when I click next, I need to answer some gating questions. So when I fill these in, I need to choose an industry, for instance, in this case pharmaceutical. And what you'll see later on is that the cover page of that presentation will change based on the industry that I selected. There are some other things that will change the document as well. I need to select the start date, because based on that some calculations are done in the document as well. I need to select which services are provided to this prospective customer. And then I need to select an engagement partner. What I do now is what happens in the background is that it takes that template and combines that with the data and then produces that document for me. So you see a presentation with the right background because it's pharmaceutical. You see that it says Jane Doe here and the members of management of NEMO, because that's what I just filled in. And it provides the signature of Jordan with these details. If I then go to this one, for instance, so what you see here is a timeline and the timeline is calculated based on the start date of the engagement. So I filled in March and based on that it calculates the rest of the timeline. Then what you see here is the scope. So I selected two services that are in scope, and based on that, these slides are inserted. Alright. That's about our compilation engine that is available, and you saw me doing that as an end user right now. What I want to do instead is not do this as an end user, but do this in an automated way, but that we have an API. So let me quickly show you what that in theory or conceptually looks like. So what we have here is our API framework. And the API gets triggered by something and it can be something rules based, maybe something in Power Automate, for instance, but another solution, CPQ, can also trigger us. And that trigger comes with the payload and that payload contains the information that we need to generate the documents. So it contains the email address of the user. It contains some data that can be just a subject like we saw before, or the name of that company, the services included. That is all contained here in payload. That is sent to the API Simplify. We take that in and we combine that with the template that has been configured to receive that information. And then we return that message with a response and that response contains a URL for that application to download to fetch that document from. So that's how these things that we will show you in just a second under the hood work. If you think about how we turn data into a document, these two things come together. So, document you created first, that was a user creating it manually themselves. This one that you're showing now is triggered. So, all of that that you've done manually would be automatically generated and Exactly. Then just Okay. For that automation to work, we need answers to the same questions. So, what I did manually is something that is contained in this payload. So, we expect that payload to be complete in that sense, containing the answers to the gating questions. We then turn that into a complete document. We're going to see that in a moment, but I guess, obviously referring back to the use case library that's going be sent out afterwards. Give the audience an example of when this would be used manually and when it would be then automated? So, you mean that, I do it in PowerPoint versus when this is used? Well, I mean, for a specific document type, so obviously you give the option to create that manually, which you've just done. When would there be an example where this needs to be automated? Yeah, okay. So, in this situation where, I do it manually, I apparently have the knowledge. I need to pull that from my brain to answer the gating questions. Ideally, that data lives somewhere already. And it can be that as in CRM or in a CPQ kind of solution It can also be that it's produced by an agent. So it can be that that RFP that we discussed earlier is uploaded to an agent. That agent did the analysis and it kind of constructed the answers to gating questions with its logic and then sends it over to us. That's kind of what we're after here. Perfect. That makes sense. So the user, instead of having to make those choices themselves, the choices are made based on data. And that reduces error and risk and stuff like that. So, yeah, there's a lot of potential and, advantage of doing that. Perfect. Well, thank you for framing it. And are we going to take a look? Yes, we are. Let's move to our environment and I will go into the skin of a sales rep at a company called B Company. And let me share the screen. So what we're currently looking at is just a blue Windows desktop, but what I'm to show you, what Stefan just showed is, what if this trigger point is an AI agent? And there are many ways to do this in agents. We have multiple platforms, think of indeed Copilot, John already mentioned Power Automate, but also Gemini environments, clean environments as was mentioned, but as we know a lot of you have invested in Co Pilots and work in the Microsoft ecosystem. So let's take that as the first example today and before showing how this is set up, configured in what we call Copilot Studio, I will just give you a look into how it could look for the sales rep. And his name is Rick and Rick has Co Pilots. As you can see from many, I think familiar, here we have a Co Pilot canvas, could be also other canvas of Co Pilots in PowerPoint or other places. But let's focus on this one. As you can see, in Copilot you have the ability as an organization to deploy agents that are specific to specific use cases. And next to a few things that are deployed by Microsoft. And in this environment, we have deployed a deals agent, which is basically within the company made agent that helps the sales reps to do, make some sales collateral. And in this case, I'm working with this company called Skybox Emeralds and I just had a good call, but before we go into all the details, we want to first have an NDA with them to make sure that we do not disclose anything towards each other to the outside world. So, I'm in chat box and as you know, a lot of people are currently just into chat. Everything they do in a daily work is not like clicking through screens, but they just want to chat with their agent. So please create me a NDA for my opportunity with Skyfort Semioles. I just give the AI agent an instruction and let's see what it does. Well, here you see it already, it understands my intent. It says, hey, you created an NDA for the opportunity of Skype for SAML. Here's a summary of what I'm gonna do. It also asks few questions and it relates a bit back to what Stefan just showed, because you see there are a few elements it needs to know, it seems. The document type, in this case an account name to fetch the right opportunity from my Dynamics CRM system in this case, the date and of course the subject. It asked me for confirmation, as you know from many agents, they also want to be sure, just like real humans. And here I give the confirmation and let's see what happens. So in the background, this agent, which is configured with a lot of rules that are just set out by my company, it will create a document it seems. Scroll there. As you can see here, it already saved it to my OneDrive, it says, but also interesting to notice here is that it now knows about addresses, where customer lives, who the primary contact is. And that's interesting because the AI was capable of fetching this from the data sources that I have throughout my company. So the user didn't have to think of that. It's just used from the backend. And one other thing that happens also, it already was nicely saved in my SharePoint, and in this case it was my OneDrive, and as you can see here, the document is nicely populated with an on brand logo, my information is populated, subject is set, references are set. So all these things that need to be in place are set and the only thing I have to do is maybe bring it to the next step to get it signed or just send it out through email. That's one example, but maybe there's more in this engagement. I also need an engagement letter after the NDA. So let's do that too. What I really like about this, as you're typing, is the simplicity of the prompt. I think one of the things that Copilot has struggled with, as a lot of AI tools have, output is always dependent on how good the prompt is. In the examples you're showing, the prompts are extremely simple. Yes. And even having typos and all kinds of things or being very short and non descriptive, it still knows what to do. So, what you just saw happening again, I just said, just give me also an engagement letter for them. And that was already enough to engage with this letter and create another document in a matter of seconds. What I've done is in PowerPoint or Word, I had to go, browse for templates, open it, go through screens, answering questions. Now just with a lazy prompt, I just got to this outcome. And here you see the document is nicely populated with all the things I needed. Certain areas were ingested by AI generated content. Other areas were just as is, as you can see here, AI content is nicely tagged so the user can still review it. So that's the current magic that the agent did for me as a sales rep, Rick. Now next to that is of course interesting and I know we have a lot of IT audience today. How did it happen? What just happened here? So I want to switch a little bit now to the more technical part and for those already familiar with, we have this tool called Copilot Studio. Copilot Studio is basically the workflow builder environment to make agents in your Microsoft ecosystem. And also already out of the box, Microsoft, as you can see here, you see a big list of agents that come with deploying CoPilot Studio in your environment, but you can also make your own agents. And I will just going to walk you through the agent that I just used for Rick to create his sales collateral. So here I have it. It's the deal documents agent and just to go for the overview here you see it's just an element that you can make, no coding needed, where you set a description so that users and the AI understand what's the purpose of the agent. You give it a model that it will use to create the contents, to ask the questions, to conversate with the user, and then you see a few stats on it, but there here is the important part. And again, maybe simplified for more complex or use cases more as needed, but it needs an instruction. And here we made an instruction for a few document types. As you can see, you are a deal document agent, you are an AI assistant that helps the company, just as I explained, to create collateral for deals. Also told it to be a company like one of the big three or four, and then a few instructions and what is needed to gather when talking to the user. So that's one element of it. Then under the hood, because my screen is not all visible, but you can connect all kinds of things, other tools, agents, topics, activities to this agent. And in this case, I connected a tool over here that's called customer documents. Basically this is a reference to a workflow where the actual magic is happening and I'm going to show you that and that looks like this. So what I have over here is a flow made in Copilot Studio, but also could be Power Automate. I will show you here the same one also. Just click on it and go to edit. So for those who are not yet into Copilot Studio, also here is a familiar flow for you to look at and you see a few steps. And the first step, of course, is this instruction that I just showed you, the deal document agent, comes generated by AI with help of the user prompts with a little bit of information. That information flows into this flow. And that's the trigger here that reference also the overview that Stefan just showed. So there the conversation of the AI agent goes into this flow. And as I told, when Rick was creating these letters, it was using information from a CRM. And for those familiar with the Microsoft ecosystem, that's all gathered in something called Dataverse. Dataverse is the sort of data layer between Dynamics CRM and some other elements in Microsoft, and it holds all this information. And in this case, I can click on it and you'll see it's about accounts. So you had to describe a similar account and there it just selects a few elements, address information just for the purpose of this example. And that's using some elements of the trigger that came from the AI. So, a lot of tech stuff, but still I think low code, you do not need to have an engineering degree to do all these kinds of things. But this is the way how you click and configure these steps in the flow. Then comes the magic. If you collect all this information and here I just have one list row item, but you could collect from several data sources and make a full payload, as you might remember from the overview, that goes to Templify to populate this pre made branded document. And that happens over here. So here we have this small Templify element. And this is the place where certain things go in, so we tell Templify what kind of assets it should be. That's a reference to the template, which user is doing, and a lot of magic JSON, the payloads, as we also have in the other one that goes into the documents. Then it knows which template to fetch, knows which data should go into it and then generates a document. And what's interesting to know is there are some other elements in this workflow builder for Microsoft. I think there are thousands even of Connector or Blego blocks or whatever we want to call them, but only a little to none except for Templify that for example are able to make a proper PowerPoint. But next to that also Word documents as in my example. So the magic here that happens and triggered by the AI agent goes into Templify, it creates the document either in native format like PowerPoint, Word or Excel for that matter. And if you want, you can also create a PDF output, then it uploads and it creates a link in my example, but that could also be signing application that you have in house where you bring it to or your email application. That's just up to you. So you are in control of what happens here, but you can, we can assure you that this middle step is always a branded and governance document that you can create with Templify, following the rules that normally would not have been followed if people just create some content with AI. So, I think that's an important thing to mention. I really like this, Ian, because one of the things you touched upon there was the deliverable, the output is PowerPoint, and that is something that we know can be an issue and it is part of that, you know, the enterprise AI gap that we've talked about. PowerPoint is such a big one. Why is that? Yeah, so, everybody and most applications can create a Word document or just a text file with some headings and with some coloring in it, but when you really want to be unique towards your customers in your collateral, PowerPoints can make these high fidelity presentations that will make you shine at your customers. And there's technology behind that, it's hard to accomplish, and therefore a lot of people struggle with PowerPoints to get this high fidelity out of automated flows. And I think with Sempify, we do a pretty good, to maybe even say the best job at that point. So, you want to get a real return on your investment, not only Colobytes, with performing sales documents, I think this combination is gold. That's a really good point. Yeah, also think about two things actually. One is, there's a lot of localization usually happening. Think about legal rules, disclaimers, classification, stuff like that. That's quite hard to do that with with AI. But also think about the process after the creation. If you need to review something that has been completely generated with AI, it takes you more work than if you know what sections have been created with AI, and the rest can be trusted in a certain way. So just trusting AI is not the most efficient way of working. It's a beautiful the blend can be a beautiful thing to have. One of the things you touched upon, Ian, was that return on investment. Something obviously I work in customer success and speak with our enterprise customers every day and they do have CoPilot and some of the frustrations that we've heard about in the industry is that that is difficult sometimes to quantify the return on investment from a tool that does so many things. And I think having a tool like Copilot in combination with Templify, you can then attach it. For example, an engagement layer as you showed there. Attaching the success of a document or an increase in revenue to documents generated in Copilot only helps the conversation that you're justifying the price tag from Copilot. I guess I know the audience. There's some technical people in the audience and I know that there's a lot of people that aren't so technical and also are mainly focused on the documentation. A recommendation is if you want to introduce us to your Co Pilot team people that manage your documents, people that manage Power Automate we'll sit down with them and take them through how we can solve their challenges and what gaps they may have that we can fill. So, excited about that. You'll know who your Co Pilot people are internally. Awesome! And so CoPilot's not the only fish in the sea. There are other tools available and I know we're gonna have a send out with a couple as well. But what other examples of tools that we can automate workflows are there? One of the other examples is Google Cloud Gemini setup next to Glean, which is also an environment where we can connect the AI agent with Templify. But maybe I can jump into one of them and have a look at this case, the one made by, powered by Gemini, made by Google Cloud. Also Google has a whole tool set to create or even different tool sets to create agentic workflows and here we did an implementation for the same dual document agents. You see the same hello and welcome to the company and let's get back into the skin of Rick again. We just made an NDA, we had the engagement letter, now they signed it and just want to create an onboarding deck. And let's ask, just signed. I have a onboarding deck? Please make a PDF. Well, let's see what it comes up with. Again, it's so simple, the prompt that you're putting in. Yes. And again, this is what I like. I think as a user it's always a bit unclear in certain AI tools what you're gonna get, but it nicely confirms what is gonna happen. Again, it goes through these elements that are needed to create this business document. In this case, again, document type, account name, etcetera. And it took also my ask about PDF, but still it asked for some confirmation. So, yes please. And then I expect it to make a document, which it will probably do. Probably? No, for sure. And I noticed that you're using this on the web browser as well, not that's when it needs to be heavy installation, like some of the solutions? No, it's depending a bit on your tech stack, But in this case, this agent was made as a web app, which you can distribute in that way to users. And again, it's up to you, as folks within your respective enterprises and companies, to choose your agent builder of choice, and that also then results in where it will pop up. So Copilot will pop up in Copilot agents, And if you do it in other platforms, will pop up another experience. And this is an example indeed of web. Well, it created the onboarding deck that also nicely summarizes a bit that it did so. Now I open it and go into the document and what you can see over here, I just have a nice onboarding deck with all the elements that were rules based or dynamic are nicely populated. Agenda is set. We have a team building for this one and here you can see all the elements. Some of these elements, again, I said also the Word documents are configured based upon rules, other elements could be ingested by AI, but here you see just a high fidelity, nice, with a timeline, all the things that it needs. So think of that, think of that you have maybe other tools to do this, but the only components that you probably will miss, as we heard with many of you, is this filling this gap between the AI, its outputs getting into a high performing business document. I think we can help you with that. And as we just saw, it was so far with Copilot Studio and with Google Cloud, which as I said, also Green is one, and probably there are more. We have this API that you can call from all these agentic low code agent builders and generate the documents as I just showed you, or even more complex ones. One thing I really like about the examples you've shown is that the output almost the same every time. And that's something I know that AI struggles with. You might give the same prompt 10 times and get a different output. You could do that again GEMINO in workflow. And while the content would change, the output is consistent. And that's what you talked about the rules part. Maybe just touch upon a little bit more about why that's so important. Yeah, so users are, lazy is maybe the good word, they just want to quickly get to an output. And that's in AI world results to different outputs. So every time you click it will probably create some other output. But what we have controlled here in this layer of Templify is that it knows the structure it needs to get as an input to make the documents. And since we have configured all these rules in these templates, it knows what it needs and will not start creating this document until it has this information elements. And then it can always create the same consistent output that would not have been possible by just using AI. So, bridging the gap between loose conversational chat and structure that you need to make this performing business document, I think that's the gap we fill. Yeah, and I think it's one of the ones that, again, I hear about the most is that we use Copilot, but we can't use it for this because of exactly what you just said: the consistency, the randomness, the fact that it may pull some of your hex colors, but they'll put them in the wrong layouts. And we work with some of the largest organizations in the world. Their brand is their business. So, the output needs to be consistent every time. And I think that we talked a lot at the start about that enterprise AI gap having that guardrail, all of that information just funneling through the Templafy templates at the end is that extra level on top. And obviously, in a lot of cases, it could just be the missing piece. So, a really good example. One of the areas, again from talking to customers where, the enterprise gap is quite wide is when actually saving sales folks and individuals time on documents, for example, internal documents that they wouldn't usually need to be so outwardly facing, but still need to pull data from a lot of systems in order to make them more efficient. And I know you've got a good example from one of our faithful partners to show. Yeah. We work with Gleam from time to time. They do enterprise search. They have over 100 connectors connecting with all kinds of applications Salesforce, Slack, Gong, and they are able to pull data from there and show that to an end user in a chat interface. And what they created is a Winwire example. So they have a document that is created by by an end user, and it shows a couple of data points that they are able to pull from these multiple sources. So there's not one source that comes from, but they have to pull it from Salesforce and from Gong or Slack, maybe. And that's work for an end user. So what they do is just ask Glean, in this case, to to build that out. So. What you see here is that template that has been configured by an admin. So you see some fields, some placeholders here, the customer name and some other fields. And then we go to Gleam, and there are two questions that need to be answered by the end user. It's the Salesforce opportunity and the customer name. And based on that, that agent can do the work. So let's click run agent to see what that then looks like. So what it now does is pull that data from all these different sources, so it knows that it needs to find that customer. And this is what is happening behind the scenes. So you see that there's a flow kicked off. It pulls the data from Salesforce, pulls the customer name, and it uses that to find in Slack and Gong what it knows about that customer. And then once it found that, there's a little bit of a prompt behind it. So it needs to figure out what to do with that data that it gathered. All that data is then collected and in a structured way, sent over to Templify at some point, because we need a structured way to be able to return a document. So these are the instructions that that agent got to do so. What you see here now is that it collected all that data. So it went into Slack and it went to into call and it collected the data that we need to create the document with. And now we ask Templafy to actually create the document so that data is now sent over. And what we now do is turn that data into a generated document. Let's have a look. There you go. So this data comes from multiple sources and is nicely put in one document. I love it. I think looking at this, this is where I could see it saving me so much time. In customer success, we have to do quarterly reporting for our leadership and I do have to go through different systems. I look in Salesforce to see renewal dates opportunities. I look at renewal health assessments etc. And I will go through Gong to see some of the key quotes. This could save me so much time. But again, it's those documents like the internal ones that are often forgot about when it comes to using Co Pilot Clean and others. So, for everyone at home, please do get in touch with us if you've experienced any of the similar challenges that we've discussed already. But anything to discuss that we haven't already brought up? I had something, yeah, but I lost it. While you're thinking about that and retracing your steps, any final points to discuss from you, Owen? You got it! I have it! So, it's not about AI, as in AI is beautiful, but AI is a mechanism to make something more efficient. So, AI can be a beautiful trigger for a more standardized process. And the reason why I like the technology that we're offering here today so much is that it helps to make an AI driven process quite efficient because it lands into a standard standardized output. It avoids the hefty reviews that you need to go through if something is randomly generated with AI. So I think what may be something to kind of get back is not think about AI as a way to make something more efficient rather than a goal in itself. And we are happy to kind of embed ourselves into that to support that. Yeah, one thing to mention, I forgot it, I think. So, after today's session, you will get a lot of collateral, which explains all kinds of documents types to use, but what I just wanted to mention, also important to know for those who can't wait for that, what I just showed you is already, if you have Copilot Studio and Templafy, it's already there. So this connector that we use is available in your Copilot Studio environment as a connector that you can just already leverage today if you have the right setup on both sides and start to test drive. Good point. And if you don't have both, that's where you get in touch with us, because whether you have Templafy and you want to supercharge some of your workflows using Gemini, Glean, Copilot get in touch If you have Copilot and you're not a Templafy customer, I think you already see the power of what can happen but essentially get in touch. Get in touch, get in touch and we can find a solution for you You may even end up talking to Stephan. So as I mentioned, and thank you for your questions I can see they've been answered in the chat by the team. But thank you so much for joining today. We really hope you found this session not only entertaining but also insightful. Introduce us to your co pilot team. I think we definitely, even if we don't find a specific solution, think we will actually have some interesting conversations to have. And again, from that brand perspective, this could be the missing piece and filling that enterprise AI gap. So from myself, from Ian, from Stefan, everyone behind the scenes at TEMPLIFY, Thank you for joining TEMPLIFY Talks and we'll see you on the next one Thank you