Alright.

Here we are. We're live in studio broadcasting to you from the lovely Columbus, Ohio. Stephen, welcome. How are you, my friend?

Thank you. Good.

Good. Good. Yeah. We were we were just joking before this started. We don't know how this was approved or how you and I were allowed on, but here we are. Here we are.

We've bribed all the right people at Tray to get on here.

That's right. But, no, I do think that this is going to be a fun session. We're talking about a topic that I'm sure many of you have heard a lot about over the last few months in your ear, AI and agents.

My name is Nate. I'm part of the GTM team here at Tray. I've actually been at Tray for almost eight years now, and this is not a joke. I actually did have hair when I started, at Tray. It's it's it's since gone. But not to be outdone by my partner in crime here, Stephen, I think you actually were one of the very first Tray users, if I'm not mistaken.

Yeah. So, like, I've used Tray for, I think, over eight years. Me for sure over seven years. But, yeah, like, I was an early adopter back at DexYP, and then that turned into Thryv.

Adopted it there, adopted at Firemon, adopted it at SaaScend, and then and then now I work here, which is weird. So, yeah, it's been quite the journey.

And not just Tray. You've used a lot of other automation.

Oh, yeah. Yeah.

Yeah. So, you're certainly one of those go to people I've had, in my corner over the years for all of the cool new things. And, if folks aren't following Stephen on LinkedIn, he's been doing a really phenomenal job, empowering the community with a lot of the agents, in content that he's been pushing out. So give Stephen a follow.

But, Stephen and I are here today. We've been on a bit of a world tour here lately. I think we've been to London, New York, San Francisco. I was in Detroit.

And most recently, we were in Las Vegas last week at the Gartner App Summit, which was a really great event. Gartner did a really good job of getting a lot of senior level executives together. And, you know, a lot of those conversations were rooted in AI and agents. And, Stephen and myself, the team there, we spoke to over one thousand different contacts at the event.

And, Stephen and our CEO, Rich, they were able to do a theater presentation where we went through, an agent demonstration of building an entree, showing how it actually works in production, and it got really, really strong reviews. And so what we thought we would do today is, bring that to the masses here, open it up to, to the community, and take you through a very similar demo. But before we jump into it, you know, I think just in our travels here of late, it's been really fun over the last, you know, two years because AI is new for everybody. And I think the big difference that, you know, I've just seen in the community is a lot of people have genuine interest in learning, you know, what's going on in the market?

What are people using?

How are people adopting AI? What are the specific use cases people have deployed to production? Because this is new for everyone. Right?

This isn't something that has been deployed for decades, and we're just rinsing and repeating it. This is brand new for everyone. And so, those conversations at the App Summit and some of the other events that we've been to over the last few weeks have been really engaging. Right?

We've gone really deep with people in those conversations. And, Stephen, you and I, we worked the booth. We were grinding it out for, yep, three days there in Vegas.

Sore feet, you know, tired legs. I think you ended up, perhaps bringing home, COVID, perhaps.

Yeah. Yeah. No. If anyone's wondering why my voice sounds the way that it is, it's because, yeah, I got the flu or something.

It's actually not COVID. I took a COVID test. It's not COVID. So I think it's a good old cold or the flu or something.

So yeah. We joke. This is your Michael Jordan flu game today. You're still showing up for the people, ready to show them what's possible with Tray and agents. So we appreciate you for that.

Thank you. Thank you.

But I thought we could start just having a casual conversation about this. You know, I'd be curious, Stephen. You chat with a lot of folks. You know, what stood out to you at the event? What type of conversations were you having there?

Yeah. You know, during Rich and I's presentation, before I even dove into the agent builder or anything like that, I asked the question, like, how many people in the audience? And there were several hundred people in front of me. And I asked, like, how many people here have built an agent?

And there was probably, like, of three hundred some odd people, fifteen hands Yeah.

Raised, which really surprised me. It actually threw me off. I'm like, of all the events that we went to where I would expect, you know, engineers and IT people and people who are at the forefront of the technology, I would imagine thirty, forty, fifty percent of the people to be raising their hands. And it shocked me just how few people.

Right? And then even people who are coming to the booth and stuff were saying, like, oh, like, I've been looking into this and stuff, but, like, not not a lot of people who actually have, like, built them from the ground up. So yeah, that surprised me.

Yeah. That's funny. I actually, after the event, I wrote down some notes. I try to just, like, capture what some of those conversations were and themes were throughout the event. And what I wrote down was an undercurrent of anxiety.

A lot of the people that we were chatting with, pretty consistently, folks are coming up, and there's this sentiment that they're behind in their AI adoption curve, which, you know, I I think a lot of that comes from, you know, every time you you open up LinkedIn or TechCrunch or turn on the news, you hear about AI agents, you know, companies, I think, are doing a really good job of using some of these production, you know, quote, unquote deployments to help out their stock prices a bit.

Sure.

Talking about the productivity gains. But, yeah, there is certainly this anxiety around, you know, we're behind, and we need to catch up to our peers. And, you know, that that started a transition, and certain folks are at different parts of their maturity curve, of course, in terms of AI adoption. But I would say the one thing I was hearing a lot fairly consistently, and I don't know if you're hearing the same, was it it feels like a lot of people are turning on tools like co Microsoft Copilot, Google Gemini Mhmm.

Oak AI enterprise, or even chat tools like Glean. And that's kind of the easy button right now. Right? It's like, okay.

We've got AI now. And, you know, for in my opinion, that's just the very beginning of the journey. Right? That's just the tip of the spear.

That should be table stakes at this point, turning some of those chat assistants on. I don't know if you're seeing similar trends.

Yeah. It's like, let's turn it on. Let's play with it. They see some success, and then it's like, well, then Yeah.

Right? Like like, okay. It's great at doing this one little thing, but we got, you know, this department over here with this use case and and this team with this use case, and then we have, like, all these other technologies that we need to, like, plug into and stuff. And it's like, okay.

Great. Do we get another, like, point solution and another point solution and another point solution? And I think that's where the anxiety is coming from too because you got the you know, your board and your senior leadership saying, AI, AI, AI. Yeah.

You know, and and that's kinda like, you know, Wall Street screaming at you, and then you got Main Street, the people who are just, like, trying to implement this stuff and and and just you know, it's it's it's a lot to take in.

Yeah. That was also entertaining. A lot of the folks who came up to the booth were the ones who have been tasked with AI adoption. And a lot of them would kinda laugh, like, hey.

What are you guys doing in terms of AI? It's like, well, my CIO told me or CEO told us this year we need an AI strategy. And it's like, okay. Like, you're starting from, you know, square one.

Where do we go? And, so there was like I mentioned, the conversations were really great, and it's fun because people are genuinely curious. Right? People really are trying to learn, you know, how to adopt this.

And, yeah, that chat use case that you're talking about, I think that was all of our first real experience with an AI application. Right? Everybody went into ChatGPT in 2023 and, you know, saw the power, and I think it was pretty clear how it's gonna impact our personal and professional lives.

But yeah. I feel like that's still where a lot of people are, and and II wrote down, you know, this this this other point where what those chatbots are really good at are telling you an answer. Right? They can go find an answer and some are better than others.

Right? But, you know, what the area that we're focused on from a Tray perspective is getting that answer, like I mentioned, is table stakes. You should be able to do that. What do you do after you get that answer?

And so, really, what we're focused on and we're investing a lot of time is actually going and executing something based on that answer in an autonomous manner. And per your point, you know, I really do feel like that's kind of phase two of this journey is sure get the answer, but then we need to be able to act in these enterprise systems. And, you know, I think that it is currently a blocker.

Yeah. Yeah. For sure. It's and it you need something that's flexible, that works for everybody with the right guardrails, versus having maybe one solution for knowledge, one solution for taking action, and then it needs to live, you know, wherever your if it's an internal use case, wherever your employees are living, or if it's an external use case, wherever your customers are living. So you gotta have you gotta have that flexibility.

Yeah. And back to your point too, you know, if you think about the journey, it's turn on the chat bots, you know, the OpenAI's, Glean, so ons of the world. And then it's as you said, let's go turn on all of the assistance that are available in our best of breed SaaS applications. And in a funny way, it's actually pretty good for Tray and other integration providers because we're somewhat repeating the same mistakes that we made with SaaS in the early days.

Right? We all went out and bought these point SaaS solutions, and that's why tools like Tray exist is to be able to connect these disparate tools. And we're effectively doing the same thing with AI. Right?

Where we're starting to go out into the same disparate tools and just turn on the, you know, AI enabled features.

Now the issue with that is it's very siloed. Right? Those assistants will be quite siloed. And if AI is really good at anything, I say it's it's it's really good at being confidently wrong.

Right? Like, it's very good at that. And so, you know, imagine the context where you're trying to get financial data out or you're trying to get customer support data out, and all that agent has is that siloed slice of data. It's gonna be incomplete.

Right? And so that's, I think, an area where folks are struggling. And then there's this other kind of trend. You know, a lot of the major players have started to invest in new protocols like MCP, a to a, etcetera.

We're learning all these fun new acronyms, and that was one of the things I was curious about to learn at the event was I was asking a lot of folks, like, are you using MCP? Are you using A2A today? And I even went as far as going and meeting with the company who is well credited with building out that A2A protocol, and I couldn't get practical use cases. Right?

And so whilst, you know, it's exciting, and I have no doubt that their AI and agents are going to demand new protocols, it's still really early in that regard too.

And so and if you're deploying this stuff, security too.

Like, a lot of people came to the booth, and I asked them, I'm like, what's keeping you from doing this? And a lot of people were just like, it's hard finding an LLM or an AI provider that, you know, is HIPAA compliant, SOC2 compliant. You know, you have all these you know, someone breathing down your neck to deploy it. And then, you know, you're just like, well, we're in the EU. We gotta, like, have regional hosting and, you know, GDPR. Like, I had a prospect call a couple weeks ago, and they came to me, and they're just like, there's so many providers out there right now who can't do regional hosting.

If you can't do regional hosting for data Yeah.

You like, GDPR is pretty much you can't be compliant with it. So yeah.

There's a lot of quick solutions. There's a lot of point solutions. But when, you know, you're an enterprise org or even an SMB with, you know, regulations that you have to follow, some of these tools just fall apart. Like, you just can't deploy it.

Yeah. Yeah. The last trend, I wanted to bring up, and I should have done more prep for this meeting. For those of you who don't know, Stephen is a Salesforce Trailblazer, and he's had billboards of his face in San Francisco wearing the infamous Trailblazer hoodie.

But the one thing I heard very consistently was a lot of frustration with Agentforce, where, you know, I think that, you know, that that ecosystem play that Salesforce is going for where they want you to live entirely within the Salesforce ecosystem. Right? They want you on CRM, data cloud, MuleSoft, Slack, etcetera, etcetera.

And, yeah, I consistently heard we've tested out Agentforce, and it was a lot harder than we expected and have somewhat abandoned it. And you're much closer to that Salesforce community than I am. I don't know if you've heard other things.

Yeah. I mean, it's kind of a, just a lot of angst around it. It's like, oh, this is a really cool thing, but then it's, like, another thing I have to pay for. And then, like, the way that you pay for it was based off of conversation, not really conversion or anything like that. And it seemed expensive, and it was hard to predict. And then they shifted the way that the pricing worked a little bit. So, and then it, you know, like you mentioned, it's Salesforce.

So, you know, it's in the Salesforce ecosystem.

Plugging into other systems and taking actions outside of Salesforce is very difficult. So, yeah, it's like, do you want to adopt a technology that just further gives Salesforce or anyone really more leverage at renewal to, like, hey. We have your data. We have your automations.

We have your agents now. So, like, if you're a business, having that data in your, like, you know, possession versus someone else's possession, is gonna become more and more important, especially with agents. Like, if you think your data is important, your model as a business in the future is probably gonna become more important than necessarily the data. Mhmm.

So yeah. I mean, it's another point solution that does decently good things within Salesforce, but then kinda falls apart when you wanna plug into other systems or tools or, you know, dare I say, migrate to another CRM, in which case you gotta build everything from scratch again.

Yeah. Yeah. Which I think is a good transition to where, you know, from a Tray perspective, a lot of the conversations that we were having with folks who are curious in learning more about agent deployments and why, you know, we're bullish about our positioning in the market from an agentic perspective is if you think about what we've built over, really, the last decade is a connectivity platform. And so one of the things that Tray does really well is we connect to systems.

And when it comes to agents, that's really important for three reasons, in my opinion. One is getting the knowledge. Right? So being able to go into those systems like Google Drive or Box or Notion or wherever your knowledge store is and accessing that knowledge base.

Secondarily is connecting to the LLMs. Right? So, one thing that is very consistent is everybody is using multiple LLMs right now. Right?

It took us probably a decade or more to get to a multi-cloud environment, right, where people are running, you know, two or three different hyperscalers.

But most people started multi-LLM, right, where it's very common to have you know, they have a deployment of OpenAI, Anthropic, and maybe they've tuned some, you know, open source model, along the way. And so Tray also allows you to bring in any LMM that you're that you want to use. Right? And perhaps you wanna use different LLMs for different agents, and for us, that's really no problem.

But then the third piece and the most important piece is providing these agents with the tools to be able to act in those systems. Right? And I think that layer, a lot of people are still, that's where I think are falling behind right now is I think it's pretty easy to get to that knowledge use case, right, and be able to answer questions with some in some of your enterprise data, of course, But then being able to take actions, like updating tickets, updating a field in your CRM, you know, requesting time off. Right?

All of those things that we do as employees, on a day-to-day basis, that's where the Tray agents really start to differentiate. So, Stephen, I know you've spent a lot of time with our customers deploying these agents already. I know we have several really great customer examples, but you yourself have been deploying a lot of agents. Are you seeing any consistent use cases that seem to be bubbling to the top right now or at least a good place for people to get started?

Yeah. You know, I have a couple that I can share here. Yeah. One I made for Gartner, so I can highlight that one. And then another, sales ops, maybe use case too. And for anyone hearing any banging on on Nate's side, he's got some house construction.

Can you hear it? Oh, no.

A little bit. A little bit. A little bit.

Supposed to be muted.

No. You're good.

At least the electricity is on still. Yeah. But, Yeah. Let's do that. I think that would be cool. We certainly wanna be able to show people some product here. So why don't you take it away?

Sure. Yeah. Alright. Alright. Alright. Cool. Can you see my screen?

Yes. We can.

Alright. So this was the one that I did at Gartner's. Right? So the use case here is, just to kinda set the stage.

You know, every organization has an IT team, and you have tickets that are coming into the IT team for either internal tickets, you know, things like, you know, my computer won't turn on or something like that, or external tickets coming in from your customers.

And there's a lot of things that come in regularly and often where a ticket is so similar to another ticket, and you already have internal documentation on it. So why not train an agent on your SOP documentation of just common issues, and have that be your first line of defense before escalating to something like a human? So this is just an example, and you can deploy this agent anywhere. Right?

So you could pick up emails, and that could kinda trigger the agents to kind of manage your ticketing system. It could be deployed in something like Slack or Teams or even SMS which was, you know, the deployment channel that I used at Gartner, just because it was easy for people in the audience to scan a QR code and interface on their phone. But just to kinda show you what this could look like, on the left hand side, I have my agent. So I'm gonna just gonna chat with it here.

And then on the right hand side, I've got my Jira board here live. So if I come in here and say, help, my computer won't turn on.

What's happening here is this request is being sent off to my agent. My agent is reviewing that internal documentation that it has of, like, common use cases, common issues that that might happen. And then, what it's gonna do is it's gonna create a ticket here on my Jira board. So here my computer won't turn on, troubleshooting needed, and then it's gonna come back to the end user with solutions.

Right? What are the what are the solutions that that we need to do to, like, walk through the person? So here are things like hold the power button for ten seconds, unplug all external devices, remove the battery.

You know? And then, what I can do is also you know, the last step here is gather additional information. Right? So, like, computer make model, when it last worked. So I can come in here. It', a MacBook Pro.

I like that the AI model just recommended to turn it on and off.

That's, well, it's actually, well, I fed it some the documentation.

Right? So it has some grounded context of, like, what I would recommend that we do. But yeah. So, MacBook Pro 2025, and I can do, like, last works Friday or something like that.

Yeah. This use case, Stephen, we did a survey a couple months back where I think we surveyed around a thousand or so different IT leaders. And we call this an ITSM use case and IT service desk management use case, and this actually was the most popular use case that people are pulling from an agentic perspective right now.

Yeah. Yeah. So here, it provided some additional context.

So then here, if I click into the ticket, you can see it added the comment here, for the details that I gave it. So, you know, then now it can be maybe be picked up by a human or let's say it solved my problem and, you know, whether it was actually the battery, I pulled it out, put it back in, and I can come in here and say, got it working.

I think I'm all set.

So this is a flag to say, hey. You know what? This ticket is complete. We don't actually have to escalate it to a human. So it moved it to the done status. So all of this happens without me needing to bring in a human.

You know, so, like, the first line of defense being AI, it can be a really powerful time saver to IT teams as they think about maybe moving into the next generation of, like, you know, deployments and managing both internal cases and tickets, as well as external.

Yeah. And just to double down on this, you know, like we were talking about earlier, those chatbots probably would have been able to do that first part of say, hey. Like, turn your computer on and off or, you know, here's the manual. That part's pretty easy to do.

But, where Tray really starts to differentiate, like we were talking about earlier, is that action piece. So going into Jira, creating the ticket, etcetera. And, frankly, that's pretty simple. Right?

Like, this is one of the more simple use cases that we have. We have agents like this who can actually triage, like, a password reset. And based off of the individual's role, and the governance within the organization can go out and actually reset someone's password or provision them access to a system that they should or should not have access to. Right?

So Sure. There's a lot of cool, you know, use cases that go beyond just updating the Jira ticket, which in itself saves quite a bit of time for people. Right? And there's a really great quote.

Notion's one of our customers, and we work with Jalal. He's their head of business systems there. And, he spoke at one of our conferences recently. And he had this really great line that I've been repeating where he said that their goal at Notion is to empower people to focus on their craft, not chores.

Right? And I think that that mindset right now where AI is is a great mindset to have. Right? Focus on your craft, not chores.

Like, updating triaging tickets in Jira for anyone who's had to do it is terrible and painful. So how much of that can we automate? Right? And there's a lot of those tasks across all parts of your organization.

Yeah. And before I move into the next use case too, I'll just pop the hood a little bit. So, like, here are the tools that this agent has the ability to do. Right? So this is where it's referencing that documentation.

And then, you know, it can search and find tickets as well as update and create them. But if, like, I dive into, like, the create one, which is what happened here, you know, it's not a terribly, complex tool as far as the functionality goes.

Let me see if I can find it.

My computer is not loading it. So, but it's, like, essentially just one step in a workflow. So it's not like a ton of different steps, kind of dangled together essentially to do a task.

And the other thing too that's worth calling out is those tools that we've given to the agent, it's not deterministic.

Right? When that prompt Stephen wrote a prompt. Right? Hey, my computer won't turn on.

What, you know, where the LLMs coming in and kind of the intelligence here and the agent is starting to play out is it's looking at the tools that it has access to and self selecting what it should use to execute said task. Right? And I think that's a really important distinction as well because a lot of, you know, quote, unquote agents that I see in the market right now are just workflows, you know, deterministic workflows, which, you know, we've been in that market for over a decade, and these are different. Right? We're giving the agent tools, and the agent itself is using its knowledge base and guidance to know which tool to use to execute said task, which is quite different than a deterministic workflow.

Yeah. And not just using which tool, but, like, also when they can start to work together. Right? So, like, you know, maybe it can look up a record, and then if it finds a record, update the record.

So, you know, there's a lookup tool and then an update tool, but there's also a create tool. So if it looks it up and it can't find it, it can then use the create tool versus the update tool. And then each tool can be leveraged more than once. So, you know, they can be leveraged in any order as and then as many times as is needed to essentially complete a task.

So, like, you know, I could come in here and say create, you know, two tickets, and it would create two tickets all within, you know, one go, you know, leveraging that one tool twice.

Yeah. I think something else that's worth calling out as well is these agents don't have to be tricked kicked off by a prompt. Right?

They could also be kicked off by a webhook or some other action taking place.

Right. Like, like, let's say you already have an intake process for tickets. The creation of the ticket could actually kick off the agent where, you know, once the ticket is created, you know, maybe you have, like, you know, a new ticket come in, then the agent picks up this new ticket, and then it processes it versus being, like, you know, chat interface or, like, an email comes in. Right? So definitely does not you know, doesn't just have to be the chat.

Yeah. So, hey, this is all well and good. ITSM use case. This looks awesome. But I'm not in IT.

Right? So this isn't that interesting to me, Stephen. You know?

Alright.

So revenue team. Let's start.

Yeah. You're on the revenue team. You're in the sales team. So let's let's maybe use a sales, maybe a sales use case here. So here's a different, completely different agent. It's my Salesforce sales ops agent here.

So let's say you wanna know, like, how many accounts, we'll just ask the question. Right? It's like a search question. So how many accounts do I have?

And then this agent has the ability to write SOQL queries just on the fly, and this agent is also object agnostic and field agnostic. So it's gonna be able to see everything in Salesforce, and it's gonna be able to, you know, take action within any record within Salesforce, which is just really, really cool. So there's thirty three accounts. You're in sales. So let's say, are there any accounts at risk? So there's a field at the account level.

I think it's called, like, a health field or something like that, and it has, like, poor, medium, or good as far as, like, a health flag.

So, I'm asking, you know, of my thirty three accounts, are any of them at risk? So it's gonna pull all those, you know, all those accounts, look at the fields, and then determine if any of them are gonna be at risk. So there's one account at risk. It's called Cypress Learnings.

The account health is poor.

And then also the renewal date is actually in here, actually a couple of days. So let me now take action and not just ask questions for this agent, and I can do something like this.

Creates a task for me to reach out to them, relate the task to the accounts and contact in that account, and then draft an email within the task for me to send next week.

So a lengthy prompt there.

Pretty lengthy prompts, but very complicated. Right? So I'm saying I wanna create a task. I wanna relate that task to the account and the contact at that account.

And then because we're talking with LLMs, I want it to draft, which I spelled incorrectly. It'll figure that out, I'm sure. I wanna draft an email within that task that I can use in my outreach next week. So here we go.

It did it. So if I try to find that account, it looks like it's here.

Here's my task.

You can see it was related to Brandon Walton, which is the owner of Cypress Learnings.

It's related to the account here, and then here is the email that, it's written for me that I can literally just copy and paste. And, like, if I gave this agent the ability to actually send emails, which I totally could do. Yeah. It could actually just handle all of this for me. But, like, it working across objects and relating things together and taking action, it is just like it's mind blowing how good this is getting. And then, of course, the email is, like, personalized. Right? It's putting in, you know, Brandon's name and whatnot, all within here.

Yeah. And you'd built another agent that we use internally, which is a pretty simple agent where after every call, it takes the transcription, it finds the key data points that we at Tray care about as a sales organization, automatically updates those fields in Salesforce, and then creates an email similar to this, which is really powerful because that probably you know, everyone hates updating Salesforce. Right?

Focus on your craft, not chores. Updating Salesforce is a chore. Right? I've never met a salesperson who loves updating the CRM.

And so, you know, subtle things like that, it probably saves thirty to forty five minutes after each call. Right? Because you don't have to go in, take your notes, update Salesforce, write the email, etcetera. So, yeah, there's some really powerful stuff.

And I know we're running at time here. And the reason why, you know, we wanted to talk about these two use cases is because, really, at Tray, when we think about the power of the platform, we talk about one platform for every agent.

What I mean by that is we have the ability to deploy agents that could span across IT, marketing, sales, support, finance, HR, etcetera. Now all the construct is going to be the same. Right? So that workflow that or the agent that Stephen just showed, that workflow being able to type something and get a response back, imagine doing that with your HRIS system.

Right? Perhaps you have Workday or something of that sort and you wanna log PTO. Right? You could do it directly from a chat interface versus having to go log in to one of those downstream systems.

And so, you know, there's really limitless use cases that you can start to build on the Tray platform, which also brings a host of other advantages around, you know, agent hierarchy, right, orchestrating agents, the ability to look at logging details, the governance, the security of it all, of being on one platform. So, we could we could talk about this for probably two more hours or so. Maybe, if we did an okay job, maybe they'll allow us to do this again, Stephen. But we have a few resources in the comments.

We launched an AI agent strategy playbook yesterday, which really made me excited for football. It's got, like, a football season feel to it. So check that out in the comments. You can download that.

And, Stephen and I are always eager to chat with folks. So, if you're on this call and have additional questions and wanna reach out to us on LinkedIn, we'd be more than happy to grab fifteen, twenty minutes and go through some more details. So with that said, Stephen, thanks for the time today. Good job showing up here.

You know, I think you did good. You powered through that. We had construction at my house, but we made it. And, hopefully, it was helpful.

So good stuff.

Yeah. I know. This was fun. We'll see if they invite us back.

We'll see. Alright, everybody. Thanks for joining us today. We'll talk to y'all soon.