We're going to talk about AI agents today. We'll go through a number of different use cases and show you what they're all about. But I'm very happy to be joined by the product manager who's in charge of AI agents. Sivan, do you want to quickly introduce yourself? Yeah. Thanks for the intro. I'm Sivan, Senior PM at Concord for three years.

As part of our AI-driven features, I'm in charge of the release and the development of the new agents feature you'll see today. Yeah. He does a great job. He wears many hats and he does some great releases. So happy to have him here. Let us know if you have any questions as we go through this. This is the first time we are ever showing the AI agents to our customer base.

So this is a brand new feature. Sivan, do you maybe want to talk about the release of this feature right now? It's still in beta, is that correct? Correct. We launched the beta a couple of weeks ago among early adopters that we have. But actually, we're planning to release this feature the beginning of March maximum. So anyone that has access to our Horizon experience will benefit from the agents feature.

Awesome. Yeah. So if you don't have access to Concord Horizon, our new platform, please feel free to reach out to myself. There's my e-mail. You can also reach out to Vanessa, our support team, etc. But yeah, get a hold of us. We're happy to talk to you about what that looks like. We will be slowly migrating all of our customers over to Horizon over the next year or so.

So that's something that you can look out for. But if you'd like to get that access now, a lot of our customers are already on Horizon, but just let us know. One other thing to plug here, Concord will be at Legal Week 2026 in New York City. I myself will personally be there the week of March 9th through the 12th. We have a booth there.

We're going to be there all week. So if you would like to get a free pass, we do have some free passes for our customers and our prospects. So we do have a few of those left. So if you're interested in attending that, let us know, and we're more than happy to get you access to the event. Let's go ahead and jump into the actual demo here, and we will talk through the new AI agents.

So the idea behind an AI agent in the system, and there's going to be a number of different agents, I think, that will be coming out in the near future. But the idea of the first agents that we have in the system are for data extraction. So as many of you know right now, let me just preface what this all means in Concord.

Right now, as many of you know, we have our AI extraction happening in the summary sheet data. So we have a number of data fields that we're going to pull out of the document by default every single time. So we're going to pull out agreement category, document types, parties, we create a quick description, we fill in our lifecycle details, and then also that total agreement value.

The idea behind the agents currently is, hey, there's probably a lot of other things you also want to pull out from your documents. We have a few customers on beta right now, and we're starting to see them pull out specific entity names like that they really need for this one use case because they just got purchased by another company.

We've also seen other customers pulling out things like liability limits or just other fields that are maybe more industry-specific to themselves. Siobhan, what else have you seen our customers using it for so far? So actually, there were different use cases. So the other form is one of the use cases definitely, but for some customers, they had different forms that were filled by different sales team members, and they wanted to track actually the answers of this form that were entirely in PDFs.

So the idea was to create a property that tracks the relevant answers to make sure that a form is compliant or not, for instance. But you can actually use that to define a specific contract type, property, fee types, service lines dedicated to your products that are not covered by our default AI extraction feature. Absolutely. Yeah. So it's basically anything that you can dream up that you want the AI to look for and extract for, you can now do with our AI agent.

So let's go ahead and jump into the actual feature here in the system. Most of you will not see this feature yet. We haven't fully released it, but this will be showing up very soon in Horizon. So I'm going to click into our new agent feature here. So it's under agents. As you can see, I currently have one agent that's been created, which is my identity clause extraction.

So that's going to be here. I want to go ahead and build a new agent now. So I'm going to show you how easy it is to walk through this process in the system. So let me pull this up here. I've already previously written this out so that I can quickly copy and paste it. But let's say for today, I want to pull a limitation liability amount for all my contracts with my vendor.

So that's all I'm going to tell the agent. So the cool thing about this is we're actually using AI, this AI chatbot to be able to help us to actually create the agent here. So right now we have this copilot. I'm going to ask it to do something, and then you'll see that it'll follow up with some questions and get some clarification before it actually builds the agent, which is really cool.

So it asked me a question here to set this up. I need to know the limitation liability amount typically appears in your contract. Is it usually a specific monetary value, or are there cases where it's described in another way? It'd be both. It's saying, OK, you gave some feedback. Does that work for you, or do you want to capture numeric amounts and leave out descriptive phrases?

Let's focus on numeric. OK, so I'm going to give it the currency here. Here we go. In your contracts, the limitation is the dollar amount. Dollar amount for now. And, Siobhan, do you maybe want to talk about the idea behind using the copilot to help build these, because I think it's a great idea. Do you want to give some background on what we're seeing here?

Yeah, actually, our intention behind that was to divide clearly how Concord must understand the specifications that you require to create a property. That's why we're taking, as Zach just did, a couple of minutes to really understand how you would like your property to appear, the format, when to leave a property blank, when to define the different states on the extraction.

So that's why you can see that the objectives on the right side are validated and very detailed to describe your intention behind the creation of that agent. Absolutely. So, yeah, it's now created this. So where we're at now, it's created this custom document field called Limitation Liability USD to capture the maximum liability in USD from your vendor contracts.

This field will extract only numeric dollar amounts and help you easily track and compare liability caps across your agreements. And I could also get more specific. Maybe I also want to see this only for specific contract types or in specific regions or contracts from a certain country. I mean, there's lots of different options here, but that's what I've done so far.

Let's see. You can change, actually, the name of the property that will solve the issue. Yeah. Because you have a property that already exists with that name. Yeah, let's go ahead and do that. Excellent. So the first step is figuring out the objectives. So the system knows, the AI knows what we're going to be looking for.

The second phase here is to actually create the document property, which is now validated, and it's a monetary amount. And so now the third phase is going to be to actually test the results in the system. So do you want to give us some background, Sylvain, of how we test and what that looks like? Correct. So actually, in the second step, as soon as a property that was suggested for creation is validated by yourself, you will be able now in the third and final step to test the extraction to make sure that Concord understood properly how you define that property and how you want the property to appear as metadata in Concord.

So here, you'll be able to select up to 10 documents that you know or not to test the extraction on this small sample of documents. So by selecting and by clicking on Bruce, you'll be able to select among the contracts that you have already in your repository, a small sample of contracts, up to 10 max, to test the extraction and see how Concord will behave.

What you'll see will just be a preview. None of the extraction that you'll see will be assigned to the document that you'll see in the third step. It will just be a preview for you to see how it will behave. So here, progressively, our agent will run in the background to extract the property you defined on the contracts that you have in your repository.

So here, we can see that some values were extracted for some contracts and for other contracts, none was extracted. And you can see now on the right side of this table, a small information button that is actually explaining why this value had been extracted for this contract. So instead of reviewing a long contract, and you can still do that by clicking on the button on the right side to access that contract.

But here, you have a summary of the reason why this property was extracted. I hope you all realize how amazing this is. For those of you that are longtime Concord users, this is a massive game changer. It means that you can actually build agents for any type of data across your entire contract database to be able to have the system pull that for you.

And not only is this going to be available for your new contracts, it's also retroactive. So you can have a pull data for any contract that's already inside of your Concord platform. So maybe there are a couple of reports that you want to create. I was at a LegalOps conference in Boulder. It was an awesome conference last week with LegalOps.com.

And we were talking about typical reporting for LegalOps teams and for legal teams. And they had all these different metrics that they wanted to be able to pull. Well, guess what? Now you can build agents to have it do that for you automatically, and then you could automatically turn that into a report. So there's just so many different things you can do with this.

This is really opening up, you know, all of the data in the system for the first time. So it's really exciting. So, Siobhan, it looks like we've got everything here that we need. We tested it. We validated. It looks good. I wanted to preview the document as well to look at more detail specifically. I could do that. That's what that other button is there.

So at the bottom here, and honestly, I don't even need to ask Siobhan because it's just walking me through how to do it on its own. But it's basically saying that it pulled these results out. It's wanting me to review them, which I've done, and asking me if I'm good with it. So I'm going to say it looks good. And if I can add a short comment here, let's assume that you are not satisfied with the format of the property that were extracted, or that you know that for some contract, the value is wrong.

So here you will be able at this stage to reiterate on the property's definition to say, hey, I'm not satisfied enough because we didn't catch this specific situation or this specific format in this contract. So you'll be able to refine the definition. So come back to step one, redefine a new property in step two, and redo an extraction in step three.

Absolutely. If you have any questions, let us know. Feel free to put those into the chat or the Q&A. We're happy to take questions. We've got Siobhan here, so this is the perfect time to ask questions. Let us know. But I'm going to go ahead and click validate and see. And it just created our agent for us. So there we go.

So Siobhan, now that we have our agent created in the system, is it going to just be automatically extracting that every time? Do we have to come in here and run a prompt? What do we do now that we have our agent? So now as the agent is configured and created, by default, the agent won't be active. So you'll have actually to approve the agent that has been created.

So on the right side, you can see all the details of the agent that has been created. So you can see three sections. The first one, configuration, is basically summarizing all the operations that Zach just done with the type, the status, and the description that you've seen at the end of the conversation. And you can see also the property that is concerned by that.

Second, which is the most interesting one here, the deployment scope. This section will help you defining the contracts on which your agents will have to run. So it can be contracts that are in a specific folder that has a certain property or that have an amount higher than a certain amount and so on and so on. Or you can select all the documents and extract this property from all the documents that I have.

So for instance, if you'd like to create a contract type property and you'd like this agent to run on all the contracts of your repository, you can simply click on add filter to define the criteria that your contracts have to match to have an extraction. That sounds good. So if we come here to activate. So actually, before activating an agent, we have to define a scope first.

So by clicking on add filter, you'll be able to select based on any property that we have in Concord, the effective date, the signature date, the creator, specific folder, or whatever the property you'd like to track. But you can also click on all documents and your agent will run on all the documents that you have in your repository.

That's pretty cool. Yeah. And I think this is really important because if you already have your data kind of segregated to certain folders, you're using our custom properties to be able to kind of designate certain documents based on document type or other things. Maybe you only want to run the agent on your vendor contracts.

Great. You can just go into that vendor contract folder and you can have a pull on that. That way you're not worried about pulling things that you don't need. So I think it's really great to be able to define that deployment. I'm going to go ahead and run it on all documents here for now. And then once we have that scope defined, you can now see that the activate button is ready to go.

So I'm going to click on activate. And the agent is going to do its thing. So, Sivan, does it immediately start going and pulling out that data or what does the actual deployment look like? So by default, if you activate an agent with a deployment scope that is defined, it will now run on all future documents that will be created on your repository.

So if you define all documents, so any documents that could be uploaded or created in Concord will have an extraction that will run on themselves. So now the question is, if you have thousands of documents already in your repository and you'd like this agent to run on those existing documents plus the future new ones, you'll just have to click on the button in deployment scope that is called run on existing documents.

So it's a one time operation that you can actually run as many times as you want, but one is largely sufficient. So by clicking on this button, you'll be able to run an agent, this agent, on all your contracts of your repository. So if we click on confirm, what will happen is that the agent will now trigger and start reading all the contracts in your repository.

So you can see in the recent activities section, the last one, that the run has started. So now it's in pending status because it's trying to analyze all the contracts that you have one by one and start the extraction. So you can see it will take a couple of minutes, depending on the number of contracts you have. The more contracts you have, the more time it will take.

But you can see by clicking on the row of this table, if you click on this, you can see that progressively, it will start listing all the contracts you have in your repository and start extracting the values from there. So you will see in this table, different columns. The first one, well, the document that is concerned.

Second one, the value that will be extracted will appear here. And in the description column, like the step three in the agent builder in the conversation, you'll see why this property was extracted for this specific contract. OK, so you can see that for some contracts, it will specify that you don't have access to this document.

So it means that if you don't have the right permissions to access a specific contract, well, you won't be able to see which contract is concerned to make sure that we follow the permissions level that we have in Concord. Yeah, that's a good point for sure. I want to make sure that that's there. And so for the ones that I don't have access to, I just won't pull anything out. Is that what happens?

Yeah, and actually it will take some time to extract the value progressively. And you can see that as soon as a value will be extracted on the right side, the last column status will appear as completed. And you'll see progressively that contract by contract that the values will be extracted. So I don't know how many contracts you have in your repository, but you can see on top of the table 100 of 1997 on the left side.

That is actually calculating progressively the number of contracts you have in your repository. It's not like instantaneous to see like if you have 10,000 documents, it won't appear 10,000 documents in once. It will analyze progressively minute by minute. And we can just let this run, right? Like I can go and do other stuff. It'll just run in the background.

We have a couple of questions here. Sorry, I didn't mean to interrupt you. Anything else you want to add? No, no, no. I can see the different questions. Can an agent be edited after it's created? Yeah, absolutely. So if we come back, yeah, by clicking on edit from this list, we can come back to the conversation and adjust the property that has been created to readjust different definitions depending on your processes.

If you add an additional contract type in your repository over time, you'll be able to adjust the definition of that property to match this new process. And the second question, if an agent can be shared with your peers? Actually, for the moment, for this first version, not yet. We give access to this feature only to admins.

So all admins in our organization can access this agent feature in terms of view rights and writing rights as well. So all admins will have the same permissions, but we're actually working on giving you the possibility to share an agent with your peers as well. But when an agent is created, a property is created as well in your repository.

So it means that you can also, with this property, filter on your contracts and also customize your view like any other properties that you have created. Awesome. I don't know. I mean, I hope you guys are excited. This is a pretty cool feature. And, Sylvain, do you also maybe want to talk about, if you have more questions, let us know and please put them in.

Do you want to talk about maybe some other potential agents that you've been thinking about or other things we could do with the agents feature? Yes. Actually, I'm thinking about two different use cases. In addition to the contract type property that you could create or any other names you would like, you can also create a new parties property because we've received some feedback on the fact that Concord by default extracts the internal party plus the external party by default.

So now what you can do is to create a specific parties property to say, hey, if you're at Concord, never extract Concord as a party from my contract, but extract only the party's name of my customers. So this is a way to create an agent as well. But second, you can also create a renewal type property to make sure that it's an auto renewal or if it's a 30-day notice.

So you can redefine any kind of default properties we have in Concord with your own definitions. Awesome. Yep. Yeah, and I know that some of you on this call probably heard us talk about like the playbooks and the AI redlining and some of those additional features. Sivan, that'll be part of probably a different feature in the system, right?

Correct. Something outside of this? For the moment, correct. The agents that are available in Concord concern extractions only. So the different types will come in a second version. So we're working on this and we'd be glad actually to understand your use case. Feel free to book a call with us so we can understand how we can solve that with agents.

Awesome. Well, this is all very exciting. Thank you. Let's see. We've got a question here. When will we get access to the agents feature? So first of all, those that are on Horizon, what do we think the release schedule looks like? So there are two ways to see this. First one, if you have access to Horizon already, we can activate the feature for you before the official release.

So you can start testing, making feedback before the release to everyone. If you don't have access to Horizon, well, we'll have to define how you'd like to access Horizon first so you can benefit from this feature. So it won't be available in the classic experience that you currently have if you don't have access to Horizon.

Yep. So if you would like to test it out and become one of our beta users until the actual release, just reach out to us and happy to get you access. It's very exciting. Awesome. I think that's all that I wanted to cover for today. Siobhan, you have anything else you want to cover? Yeah. There is one last point that is interesting to mention.

Actually, if you navigate through your documents, you will be able to see that the property, if we navigate in the documents, in the docs, in the navigation bar. Oh, just go into the docs. Okay, gotcha. To the inbox, yeah. Actually, what's interesting is that if you'd like to filter on that property or customize your view, you can see that if we look for a limitation, how was it called again?

Limitation of liability. You can see that the property is here and you can now filter on that property. That's very powerful. So you can create your own reports and you can also customize your view. If you'd like to add this specific property that is very important to track on a daily basis, you can even add that as a column in your view.

So potentially, since the time we started talking, you see that some values have been already extracted for your contracts and the agent is running as expected. It's doing a great job. I know for a fact that... For sure it won't be done. Yeah. It won't be done. Maybe it is. We can double check, but actually the agent is running as expected.

Yeah, it looks really good because we have a lot of the same documents in here because this is test data, so I can tell that it's doing its thing. These are all the same documents, so that's great. Yeah, this is a really exciting feature. It means that if customers want to be able to pull out those different data fields that are going to be important for them, they can now do that.

It's all right within the system. So very exciting times here at Concord. We're just getting started here in 2026. We have many, many other things that are coming on the roadmap, but for right now, let's focus on this one. And if you would like to get early access to this, let us know. We're happy to help you with that.

If you have questions, reach out. Any other last questions for today? Let us know real quick before we jump off. Otherwise, we will take it from there. All righty. Well, Siobhan, thank you so much for joining today. Thank you for building this awesome feature. It makes my life a whole lot easier, and I know it makes our implementation team life a whole lot easier as well, so thank you for that.