Stop Digging, Start Asking: What Connected Data Means for Operators in 2027

Hello. Hello, everyone, and welcome. We are glad you're here. It's that time of year when 2027 plans are taking shape. Budgets, goals, and priorities are on the table. And if you're like most operators, you already know the answers to your biggest questions are somewhere in your data. They're spread across reports, systems, and spreadsheets, and getting to them usually means a lot of digging. Today, we're talking about what changes when you ask just a question instead. We'll look at what connected data and AI mean for operators heading into next year, hear how one operator is putting it to work right now, and leave you with practical next steps that you can take away for planning. Before we get started, a few quick things to note. First, we are recording this session and we'll send the replay to your inbox tomorrow. Second, we want to hear from you. Drop your questions in the q and a box on the Zoom control panel at any time and our panelists will answer as many as we have time for at the end. Keep an eye out for a quick poll in just a moment. We'd love to know where you are in your data journey. Now let's meet today's presenters. First up is Julia Crawford, senior vice president of product management at Grace Hill. Julia brings more than twenty years of product leadership to the table. She actually guided PerformanceHQ from first content concept to launch, building solutions around real problems property management teams face every day. Next is Arash Rashidi, principal product manager at Grace Hill. Arash leads our data products, market intelligence, and APIs. He spent his career where machine learnings meets product from deploying deep learning models for companies like Chevron to founding his own legal tech company. Today, he brings that experience to the market intelligence behind HelloData. And finally, Eddie Mehmeda, head of data and analytics at Harbor Group International. Eddie leads Harbor Group's work using data and AI to make investment decisions faster and more consistent. Before that, he spent, six years at Blackstone where he helped build a platform that gave, real estate teams real time data on deals and portfolio performance. I wanna welcome our speakers. Thank you for being here. Julie, I'll hand it over to you. Alright. Thank you, Kayla. Hello, everyone, and thank you for joining us today. I'm gonna kick us off with a poll. It's just to really level set everyone on where we are today with AI. Okay. So how is AI showing up in your team's work today? Let me give everyone a second to answer. All right. We have about half the audience participated. Give them just ten more seconds to get your answer in. And the results are in. Right. Okay. About 41% is more broadly adopted. I mean, this is this is in line with a lot of what we're seeing. We're we're actually seeing a little bit higher in our in our policy customers today that have adopted our generative AI policy and are moving towards, more broadly adopting the use of AI. I know here at Grace Hill, our CTO, Kris has done an incredible job of quickly rolling out new cloud capability. I I cannot say enough how using cloud tags in Slack has been a game changer for us. But that said, we're moving beyond the basic best practice for AI in this webinar. I was reading a McKinsey article recently that said most organizations are still measuring the success of AI usage or adoption. But with rising costs and the overall cost of AI, we need to make sure it's delivering real value. And the goal of this webinar is to share one way to drive that value. Go ahead. For years, the most valuable people on your team are the ones who knew where the answers lived. Which report, which system, which spreadsheet someone built three years ago? I'm sure all of you can picture that person right now, and they're great. But that knowledge quietly becomes a bottleneck. When they're on vacation, buried in budget season, or they leave, think about all the times you've implemented something that maybe didn't make it into your onboarding plans. Answers slow down and decisions wait. So the skill that matters now... Oh, sorry. As AI, is connecting property data across platforms, the answers are getting much easier to reach. So the skill that matters now isn't knowing where the answer lives. It's knowing what to ask and being able to trust the data behind that answer. An AI consultant put it to one of our colleagues this way, and it stuck with all of us. We're moving from paying people to have the answers to paying people to ask the right questions, and that's what today is about. Okay. So why now? Two reasons. First, the technology moved. AI started out answering from what it already knew, then every vendor, us included, put a chat experience inside its own product, and each one only saw its slice of data. Now AI can work across your connected systems under each person's permissions to answer questions. That means your people can log in less and ask more. Whether you're using Claude or ChatGPT or Copilot, the AI will go into your systems for them. As a software company, we could see that as a threat. We don't. If your regional manager gets a better answer faster without logging in to one of our products, that's a win for her and for you. The second reason is timing. Most of you are finalizing your 2027 plans right now. The groundwork you lay in the next few months will decide how ready you are when you enable your teams to start asking questions. So now I'm gonna hand it off to Arash to get into the details of what MCP means. Thanks, Julia. So for the next few minutes, I'll explain the plumbing behind what Julia just described in plain terms. MCP stands for model context protocol. It's just it's just a standard that lets AI tools connect directly to to where the data lives. The model is the AI tool that you ask your question into... You put your question into. This could be ChatGPT, Quad, Copilot, or whatever else someone comes up with next. The context is the information that you need to answer the question. This can be from your property management system, your leasing data, or your reports. The protocol is just a rule book that all systems have agreed to. No single vendor owns it, making it universal. So the analogy I use on why the rule book is important is back in my day, Apple had their own plugs. So if your phone was dying and you didn't have access to an Apple cable, you're out of luck. It has just... It has since switched over to USB c, and now almost everything you use can be interchanged, and you can charge your cable with any cable in the room. MCP does the same thing for AI. Each system can connect once, and any AI tool can use it. Now adding a new data source stops being a custom project, and you're no longer locked into one tool. If a better one shows up next year, your connections still work. How is this different from how we get data today? Well, today getting answers means that you have to move data first. Whether it's an API or a bulk file drop, someone has to map it, load it, prep it before anyone can ask a single question. MCP flips that. Your data lives... Your data stays where it lives, and you ask AI to interact with it directly, in plain English. So how does a question actually turn into an answer? Say you ask which properties are priced above their comps but scoring lower on mystery shops. The AI has to figure out three things, which systems you have connected to, what it needs, from that tool menu in each system, and the permissions for those tools. For this question, it needs two things. So here's what it used. Mystery shop scores for, Mystery shop scores from PerformanceHQ and rent versus comp data from HelloData. That's everything that the question asked for. And here's what it didn't use and why. Training is available, but the question didn't need it. The PMS is connected, but it wasn't needed as well. And then PHQ HR records are locked, so the AI couldn't even access those if it tried. The result is one answer. The properties priced above market with the weakest shop scores and the sources behind each number. So the hardest part is not... Is no longer getting the data. It's actually knowing what to ask and how to ask it. On the left, you have a generic question that requires your AI to infer what doing means in this context. Are you referring to occupancy rates, total revenue, reviews? You'll likely get a generic summary that tells you very little. On the right, you have a sharper question. Which Southwest properties had renewal rates below 50% last quarter versus a year ago? Here, you've defined a scope, which is Southwest Properties, a metric, which is your renewal rates below 50%, a time frame, which is last quarter, and a comparison versus last year. That question narrows down what the AI needs to look for, And the sharper the question is, the more actionable the answers become. Mcp is a bridge, so it carries whatever is in your systems to the AI. There are two things to note about this. Access rules carry through, and that cuts both ways. If you set your access too tight, people get incomplete answers. And if you're too loose with it, people see data they should. Second, your data sets the ceiling. If it's missing or loosely defined, the AI may not stop and tell you. It'll give you an answer and sound confident in doing so. With the right data governance in place, MCP will allow an ops leader to compare training completion to online reviews, property by property, regional manager to connect survey scores to upcoming move outs. The sky's really the limit in the type of questions you can ask. The pattern here is that the best answers typically live between systems, and that's why you wanna connect them. Nobody knows that better than the operators doing that today. With that, Julie, I'll hand it back to you. Thanks, Suraj. Eddie, welcome, and thanks again for being here. Before we jump into the questions, I think we'd love to hear a bit about you, Harbor Group, and what you do there. Sure. Thank you. I feel very welcome. Yeah. So my name is Eddie. I've been at Harbor Group for about a year and a half now, and my primary role is is head of data and analytics and also AI enablement as sort of AI is becoming more prevalent you guys all know. But the focus here is, you know, the elevator pitch is, you know, Harbor Group is one of the largest transactors in Multifamily as of last year, and we get to see a lot of deals and we see a lot of data. And one of the benefits of our business is we're also vertically integrated operator. So everything we buy, we actually manage as well. We have a huge credit portfolio. So as you sort of start adding up all the different business units that we have, we basically have access to something on the scale of one to 1,100,000 units worth of information nationally. And and that's something that people certainly wanna glean into as they're making investment decisions or asset management decisions to kinda understand, you know, not not only how our assets are performing, but other assets in the market that we're familiar with or have looked at are performing as well. So data is becoming very lucrative, especially as these AI tools are sort of competing in the space for, know, who has the better model or or sort of what's the latest and greatest advancements in AI. We're sort of treating the the tools as a commodity, you know, whether you're Claude or or Copilot or ChatGPT. We, as an organization, started off on Copilot and then eventually started rolling out Claude to some of our highly analytical folks. So we're sort of agnostic to all of those those tools, but something that we can't replace is is our data, and that's sort of our secret sauce. And one one component of that is, you know, standard operating procedures and storing the the multitude of SOPs that we have to maintain across our 1,600 employee base across the entire nation, and and that's sort of where Grace Hill comes into play. And and the MCP connector has been, really game changing for for a lot of the folks that work on-site teams. But, yeah, happy to jump into any and all that. Well, so you've... You touched on it a little bit, but maybe specifically, what what specific problem were you trying to solve with this? And at what level? Was it was it corporate, on-site teams, or both? Yeah. I I think it's for both. And and, you know, as as it relates to Grace Hill's MCP connector, you know, Harbor Group has a ton of standard operating procedures across all the different properties that we own, all the different jurisdictions that we do business in. So California SOPs might be slightly different than New York SOPs, etcetera. And a lot of the times... And it's to no discredit to the Grace Hill prep... Platform. It's sort of a lot of the times people were logging into the system and spending a lot of time just finding the right thing, searching through all the different SOPs. And if you just so happen to be someone that works on multiple properties, you're you're kind of finding a needle in the haystack. So the problem was really, you know, people know the data that they want, and they just wanna go get it without having to sift through a bunch of stuff. So so that that was the main problem. And, you know, we we reached out to to you guys very early on sort of seeking an answer to this this problem, and and you guys came back with this MCP connector. And it's been game changing in the sense that people don't do do the digging and finding anymore. People just kinda, like, ask the question and get the direct response that they need from all the SOPs. So, you know, having joined a year and a half ago, I I had to go through all the different trainings and sort of learning all the policies, even things like, you know, what are our company's benefits, all all that kind of stuff. Just being able to use the MCP and ask questions on top of it and directly get answers has has been really the key to to all this. And I think it's just increasingly a trend across all the different software providers that we use where it's... To a rush's point, it's kinda like keeping the data where it lives, but giving people a a more meaningful way to interact with it. So what would you say is a question that someone at Harbor Group can ask now that maybe they they couldn't or wouldn't before? I'll answer this a bit differently. I don't I don't know that there's questions that they couldn't answer before. I think they could depending on how much time they really spent. Right? And and it's sort of like you could probably do just alright logging in the interface and find the right policy eventually. But but what I would say is the thing that they could do now that they couldn't before is, you know, most of our employees are using AI tools every day, whether it's to simply change the grammar in an email or reword something or or sort of give a status update on various things or even some of the transcription tools for meetings, what... Whatever have you. I think AI is sort of becoming core to how people operate here day to day. And as a result, it's never kinda just... You know, you're never just using Grace Hill on its own. Grace Hill is sort of one piece of the larger puzzle to something a site manager might deal with or something an asset manager might deal with on our side. And I'll give you an example. You know, you might have a tenant walk in or send an email and say, hey. You know, I I had a leak in my unit or or or something like that. And the the reaction for some of... Someone on the site is not, oh, let me just go respond right away. It's sort of like, read the email, go read the policies, go pull the latest policies, go pull this, tenant's unit information, go pull all of the different work orders this unit had, go kinda cobble all this information together. So Grace Hill just becomes one component of that AI workflow for them where it's like, they're no longer left in the blind of, hey. You know, here's what I think I should do, but our SOP says we should do something entirely different. Hey. We had a slip and fall. You know, someone might wanna react right away. Oh, we should do excellent. But then we have a whole policy that kinda handles slips and falls. Right? So it just changes the way people operate and makes people more effective because they can do everything from one spot. And, you know, I I gave the analogy before. It's sort of like your browser is your browser. You can still log in to all the different systems. You can still get all the data that you want, but it's sort of changing the behavior of people. You know, I imagine back in the day, you know, googling something was was sort of a verb, you know, go Google it. And and it's like... Now it's like, just go ask AI or go go ask Claude. And and having all these connectors integrated to the point where it could speak to all your different systems changes it for people where they don't have to leave the one spot they're working out of. I would imagine also that that quicker turnaround time leads to just a better a better overall resident experience also because now they're getting a response faster. Quicker response times. Yeah. Okay. So what what surprised you the most about about this? I think the thing that surprised me is sort of the way people have been using it. You know, one thing that's, super interesting is in the early days of AI, everyone was sort of coy and and hesitant to admit that they're using AI. You know? But if you looked at all the emails, it's like, all of sudden, everyone's punctuation is perfect. Everything's grammatically correct. You know you know, everything's formal. And I think we've quickly gotten past people's fear to admit that they're using AI, people started embracing it more. And once people got over that hump of actually embracing it more, people started having open and honest conversations about, hey. We could use AI to do this better or or sort of have you thought about using AI for for this use case? And I think it's also entertained a lot of these kind of conversations where we reach out to our vendors and say, hey. We'd really love some sort of opportunity to use AI here to get our data more efficiently and and vendors being more open to to that. So I would just say the thing that surprised me the most is people's reaction to this. Because I think in the beginning, when there was a lot of, like, noise and fear of it's gonna replace a bunch of people's jobs and, you know, it's gonna change the way everyone does everything. But it's it's almost been none of that, and it's been just people embracing the tool and just a new way of working. And then people being intellectually curious about, you how far can we push the envelope? How many systems can we connect so that I don't have to keep doing this manual thing? And it's just changing the way people think about problems. You talked a little bit about a new way of working. Is there anything specific there where, you know, are people working within the AI tool and not using using the actual applications anymore unless they have to really go in? I would say that's one thing. It... It's sort of like, you know, peep... People having to log in less frequently. You know, as as a as a property manager, you know, we also use a property management system like many other property managers. Right? And being able to just ask a question, who lives in tenant... Or who lives in unit, you know, one one a of this building? You know, how many renewals do I have upcoming? Sort of what what are what are my policies for doing all these different things? It it... People just get into this q and a mode. But I would say that it it also changed the dynamic of how diligent people are. Because I think a lot of the times when you're dealing with humans, people are sort of afraid to ask certain questions. You know? It's like, I don't wanna be judged. I don't want my coworker to think I don't know what I'm doing, or maybe it's too late in the game to ask. And, know, I I should have known the answer to this earlier. But I think what what this is allowing people to do is you... When you have all the source of truth in one spot and you know it's the source of truth, there's no one sort of monitoring how many questions you ask to Claude or to copilot, you know, about a certain topic. So if you're less versed in something or your context switching all day, it... It's interesting because now people spend a lot of time just kinda getting up to speed and going through the iterations and going through the back and forth chats and prompts to get familiar with the topic at hand so that when they come to a meeting or in front of a tenant or or to have a formal conversation, they're kinda more informed. And and we've noticed that across the board, people just are more informed on all the different topics that we're talking about. Makes sense. Alright. If an operator on this call wanted to start tomorrow, what is the first thing that you would tell them? I would say just use it. Just use the tools, play with it. You know, I think one of the biggest points of feedback we get is, you know, hey. Is there training material on on how best to prompt or how how best to do all these things? And I'd say, yeah, the training material is good, but it... It's sort of tough to keep up with how fast these things are moving. And I would say even in the early days of the Grace Hill MCP, people were like, what... You know, why do we need this? Why can't we just log in and and and just... It kinda works. Right? Why why do we have to change it? And then the minute that people go and use it and try it, they realize, oh, you know, why we... How how did we even operate in the old days? And I... I'd say the good first step is to just dive in. I know there's a lot of hesitation on people to, you know, measure ROI and make sure that there's a valid use case, and we have a budget for it. And I think we, as an organization, have been very fortunate in the sense that our senior leadership here has largely embraced AI and and has encouraged everyone to use AI to improve their their operational efficiency, but also not really replace the human in the middle. Right? So making sure that you're empowering your your people to actually use these tools and use it in a governed way, in a sophisticated way, and and be informed about how they're using it, connecting all the sources that are valuable to your organization, which Grace Hill is a big component of that for us. And then just letting people have at it and and letting people share their their wins and and sort of building this culture around the fact that it's okay to use AI and and just use it in a way that makes sense. Did you find anything interesting with the adoption curve across the company? Yeah. I... I'd say there's there's definitely a champion behavior. You know, when you initially roll this out across any organization, especially true for us here, We rolled it out to 1,600 employees. I'd say of those 1,600 employees, there were probably, like, 30, 40 people that were gung ho and all in and using it all the time. But those 30, 40 people go talk to their two, three friends. Right? And the next thing you know, you have a 100, 200 people using it. And they go talk to their two, three friends. And then lo and behold, you have people reaching out and saying, hey. I feel left behind. How do I get plugged in? How do I get caught up to speed? Right? So there's sort of a network effect there. And I would just say, identify your champions, invest in them, make sure they know how to use the tool, that natural growth is just... It... It's so easy to to to witness once people realize the value of it. I love your comment, Eddie, about just diving in and and and using it. I Just do it. I mentioned, Chris earlier. He's obviously one of our biggest champions. And I... You know, every time I would reach out to him and ask him a question like, hey. And, Claude, how do I do this? He'd he'd always come back to me and say, well, have you asked Claude? Have you asked Claude how to do it? And I'm like, I guess I don't... I didn't think about that. So we got this whole thing where It's like having the world's smartest person sitting right next to you the whole time. Yep. Okay. Well, thank you for sharing with us. I'm actually gonna hand it back to Arash now to talk about... A little bit about governance. Appreciate it. Thanks, Julia. So far, we've talked about what AI can do, and Adi just showed us what it looks like in action. Now let's talk about... A little bit about what best what best results you can get... How you can get best results, I'm sorry, by having the right data governance in place. This is typically where most organizations tend to get stuck. So the biggest objection we hear is we don't trust what the AI tells us. But when you dig in, it usually isn't about the AI itself. It's more so that the AI doesn't have enough business context, like what your metrics mean or what source to actually trust. The good news is that we can fix this by agreeing on definitions and ownership within your organization. This really comes down to three different buckets, defining the what, who, and the where. The first the first one is, agreeing upon a shared definition for each one of your core metrics. Second one is a named owner for each one, plus rules on who can see it and who can't. And then the third is the source of truth system, that the number comes from. Once you've made these decisions, write them down, and that's the start to your data dictionary. Here's an example of what it would look like for net effective rent. Our so... Our source of truth is HelloData. In the middle, you've... We've provided a written definition, a formula, and any assumptions that we've made. And towards the bottom, we have the owner who could see it and how often it's reviewed. The template is simple by design. When your AI sees this page, it uses a definition instead of making up its own. That means that anyone on a team can ask, what is the net effective rent across my comp set and get the same answer and be confident that it's actually correct. This may sound like a really big project, but it doesn't really have to be. You can start small. You'll likely wanna pick five to 10 numbers that your business actually runs on, to see enough of the difference. What you do is just name, name an owner for each one. Ideally, this is a person, not a committee of people. And then just write it down. I find that Google Sheets or, Excel work just fine. Just make sure it's centralized and easily accessible. Point your AI to those trusted sources first, then expand as you as you document more. That way, the answers that you that you get are ones that people can trust, and then that trust will drive additional adoption just like Eddie said. Raj, just to just to jump in here. Yeah. Go for it. You know, you bring up HelloData. One thing I I failed to mention is we we do look at roughly 6,000 deals a year, and a big part of that is being selective about what we underwrite. The the reason we've had to historically be selective is because it takes human hours and time. And one of the skills that we developed internally here is the ability to just give any list of addresses and automatically pull HelloData market comps for that. So our underwriting process also has this deeply baked in where people could just give a... You know, provide a list of addresses from something that they receive like an OM or something from a broker and and real quickly get an assessment of what does market data look like for that from from the HelloData concept, which, again, is one of those things where you could totally log in the browser, upload an Excel file, hit run, and get the the reports. But when you're doing, you know, 15, 20 other things for an underwriting model, it's just one less thing to worry about. Yeah. That's an awesome way of accelerating your deal flow. Yeah. Appreciate appreciate you hopping in there and highlighting that. Yeah. So none of this really needs new technology. It just needs a few decisions and someone to own each one of those decisions going forward. With that, I'll hand it back to Julia to bring us home. Yeah. Thanks. Thanks,. So as you're building out your 2027 plans, there are really three shifts and and things to think about and plan around. First, AI becomes where the work starts. When we were doing our our prep meeting, Eddie said something that the team's been thinking about, and I keep coming back to, which it's becoming the new browser. Your teams are gonna start in their assistant now, and that assistant's gonna reach into the systems from there for them. Second, the more systems that speak MCP, more and more, they're going to. And many property management systems like Eddie's, for example, already have their own. As more of your vendors add connectors, a single question can span them, and that's where the most valuable answers live. And then third, when you review your tech stack, don't just ask whether a vendor connects. Of course, make sure your vendor is respecting the roles and permissions you've already set in their system, but also ask how they connect and what data you can actually reach. A connector doesn't always include everything you can see in the application, so check what's really available before you can... Before you count on it. And all this to say, the bottom line is it's your data. The value we talked about today lives between the systems. So the more of your data you can bring into one question, the better the answers you get. Alright. So here's where I will leave it. In 2027, the advantage goes to organizations that give their teams access to the right data so they can ask better questions and trust the answers. The technology is here. Whether you're already, whether you're ready really depends on decisions you can start making now. We'll share the recording after today. Oh, and, one other thing to watch for is in November, we are releasing the Grace Hill PerformanceHQ MCP, that Eddie has so graciously been using for us and testing with us. But, it's gonna start with your policies, training, resident survey, and mystery shop data with more capability to follow. So, I'm gonna hand it off to Kayla now, and let's open it up for questions. Okay. Nice job. Really great info shared. I know I learned more about MCP and how it can connect to your data and systems. We did have a few questions come in during the session, but if you haven't done so already, you still have time to submit those questions. So first, how should a smaller operator without a dedicated analytics team maybe approach something like this? Yeah. I mean, I can take that one. I think that for a smaller operator, it almost becomes your data and analytics team. I mean, it's an advantage whereas before, maybe you didn't have the capability to, to leverage something like this or you didn't have the ability to have a full doing this analysis, now you have the ability to to use your AI, basically, as your team analyzing your data. Very good. Let's see. And, Eddie, I think this one is for you. What would you do differently if you were starting over? Yeah. I mean, it's it's it's tough to answer that because sort of the rapid involvement of AI, you know, it's now hindsight's 2020, but we were kinda just rolling with the punches, especially as all these new developments came out. And I would say if if we had to start over, I would say investing more in kinda understanding what MCP type options exist from vendors and if not, starting those conversations earlier. Because a lot of the... You you know, some of the larger vendors that we work with are already well ahead of the curve, but there's also smaller vendors that we work with that haven't been as open or privy to kinda developing a way or a meaningful way for AI to interact with them. So, you know, people get a little, you know, jaded now because they use all these systems that do support MCP, and then, god forbid, they have to go now use a system that doesn't. It feels like stepping back into the stone ages. So tough to kind of look at that and say, hey. We should have done that better because all of that was new information as we were building. But now that it's sort of here and here to stay, it's it's really... And and honestly for us, it it goes into the scorecard of assessing a vendor in general for all net new vendors is, like, how ready enabled is this vendor to to connect to our AI tools? Yeah. Absolutely. It's not gonna be just a nice to have, but critical to your business. For sure. And here's one that I hadn't thought of, so thank you for submitting this. If we already get data through SFTP or APIs, do we have to give that up to use MCP? No. I see MCP as an additional layer on top of that. Really, APIs and SFTP enable workflows and enable you to move data into your warehouse and analyze it in different ways. MCP allows you direct access to that data. And, you know, I would say those those use cases are all different. Yeah. But definitely additive. Okay. Let's see. We do have a chat. We have a a question from Jake. Let's see. I'm not as worried about what I cannot connect to. I want to know what data I can connect via MCP, especially if it's data I don't want available to Claude or ChatGPT. Even if Grace Hill says that certain isn't available, how would I verify? Oh, it's a great question. So I think the first thing that you can do is look through your roles and permissions, and you could just ask... Well, you you could just ask the vendor that you're working with, let's say it's Grace Hill, for a list for their... For a list of tools that we give our MCP access to and what those permissions look like for each one of those tools. That'll that'll be a good starting point for you to analyze what is available and what isn't. And I'll just add on, like, you know, some some of these tools are, you know, eat your own dog food. And and one of the ways we use Cloud is we just say, hey. What data do you have access to? You know? What data can you access? Right? And and it... It's it's a way to just kinda introspectively look at the tool in general and understand what data is available there. So so we do that across the board in general. Yeah. That was gonna be my answer to that question is you can you can ask it. How convenient. And lastly, it looks like that, we have one more question left. You mentioned AI only sees what each person is allowed to see. So it's kind of a related question. Can you say a bit more about how that works? Can I? Do you mind reading that question again? Sorry. I apologize. Yeah. You mentioned, AI only sees what each person is allowed to see. So I think they're asking about permissions. Yes. Yeah. So, each user within the different applications or systems that you use have their own roles and permissions. Those roles and permissions get carried over. For instance, if you're an admin in Grace Hill, you have admin level roles and permissions, and you have access probably to the widest range of, workflows or features within our system. If, you know, if you have a lower access than that, that gets carried over. So it's essentially a one to one. Where you really wanna take a look at is what tools you have available through MCP. You know, not all systems will share all of their features or functionalities via MCP is what we've come to found... To find, and that's what Julia had alluded to the beginning of the, presentation. So if the tool if the tool's there, you don't necessarily have to worry so worry so much about the permissions getting carried over. That's part of the protocol, for MCP itself. Very good. And with that, that wraps up your submitted questions. As we approach the end of our session, a big thank you to Arash, Eddie, Julia for sharing their insights, and to all of you for spending time with us. If we could go to the next slide. Our experts are here to help you navigate every part of property management. If you're ready to talk through how connected data could support your portfolio, we'd love to connect. Reach out to us at Grace Hill dot com or reply to that follow-up email that's going to hit your inbox tomorrow with the recording to connect to with our team. You'll also find our presenters on LinkedIn using the QR codes on your screen. You can reach out and keep that conversation going. Thanks again for joining us. We hope you have a great rest of your day. Goodbye.