Skypoint Logo - AI platform for industries
Close this search box.

Integrating AI to Boost Efficiency & Chat with Your Senior Living Data

Stay up to date with the latest customer data news, expert guidance, and resources.

Skypoint was recently featured on The Raising Tech podcast to discuss AI in Senior Living. Raising Tech is powered by Parasol Alliance, The Strategic Planning & Full-Service IT Partner exclusively serving Senior Living Communities. Big thanks to Founder & CEO, Amber Bardon for hosting us.


What does it mean to chat with our data?

Now that AI is everywhere in the news and on your phone and social media. What generative AI has done is provide a means of being able to have a conversation. With a chat agent, right? And previously that chat agent used to be very deterministic, like example, you’re going to a customer service website, or, you’re on a customer service call and you’re like, pressing 1 to go to this tree or.

I have to give it my name, and it’s going through a tree of thought. What it means to chat with your data is using a large language model like ChatGPT that understands the context of your business and your business data and your data systems, whether that’s policies and procedures that have been tucked away in SharePoint or a network drive, or you have your data systems in Senior Living like Point Click Care or Yardie.

UKG, Sage, and the list goes on having an easy way to ask what is going on in the day to day operations and query these systems. That’s what I mean by chat with your data and having a natural conversation with that agent.

How does Skypoint view AI and how do you bring AI to the industry?

I think if we want to look historically, the definition of AI has changed earlier. What was AI? It was traditional machine learning models that were able to predict or help score or provide some sort of risk assessment on data. And you had to go through arduous data science and machine learning and training, and this was the days of big data.

So then AI went making predictions on: Here’s the data on what I have today and can AI help predict what’s going to happen tomorrow? So what we call predictive analytics, when people are talking about AI today, it typically sounds like they’re talking about ChatGPT or generative AI.

What is generative AI?

We saw the growth and the popularity of large language models. These large language models are the engine to having these very natural conversations. It is the means of how we work with ChatGPT. ChatGPT has large language models that enable it to have a natural conversation with us.

Knowledge Corpus

And then they’ve taken a corpus of knowledge or think of it like a library. They took the whole Internet. Fed that library of knowledge to this large language model that can have an eloquent conversation. And now, all of a sudden, I can ask questions about this library of knowledge of what’s been going on in the world. Whether it’s legal work or different domains of knowledge. I can now have a conversation and it can help me do things. That large language model isn’t just about predicting the future now. It’s about help me do work in my certain work stream. Help me write this email. Help me write this procedure.

And so that’s the big leap we’ve made is now we’re not just trying to predict the future. We are actually having AI help in our day to day tasks. So that’s what generative AI is really helping us do. It’s not just being able to analyze what’s happening. It’s chatting with your data and doing something with your data.

And I think that’s always the missing piece for people is well, okay, what can it actually do? I think that’s also what’s scaring people, right? It’s we’re saying, Oh, is it going to take my job? How am I going to be able to pivot in this day of generative AI? And the term I really love to use and what Microsoft is making more popular is this idea of a copilot.

Copilot is there to enhance what you are doing as a human and how you can do it better, faster, more efficient. And so I think this era of copilots, this era of generative AI, something is going to help, whether it’s senior living operators or beyond, either work with their data and their systems, and then in addition to that, get their work done faster.

How did Skypoint come to get into senior living?

Skypoint as a platform is a means of providing, whether it’s senior living operators or beyond, the data and AI infrastructure. Getting to the point of having a place where you can land and bring your data, unify your data and make it consumable for a large language model. So it knows what to do. And you can have that conversation on your business specific data. Again, whether it’s PDFs, audio files, system data that’s in tables and rows and columns. How do you make that consumable? You need a backend infrastructure to do that.

Senior Living’s “Leapfrog Moment”

In senior living, we’re seeing a “leapfrog moment” for operators because a lot of the operators I’m talking to lived in spreadsheets. They’ve never invested in building a data warehouse or a data lake house. Those terms typically go over their head. And so how do we allow that? How do we accelerate them to being able to have that infrastructure with clicks of a button?

How do we provide a fractional data and AI team to help them accomplish and tackle certain business use cases that are going to move the needle for them, rather than assuming every operator has the same problems and fitting them into a cookie cutter box. The way we approach it is really working in a design thinking process with the operators and saying, okay, where are the places you’re spending a lot of manual time and doing a lot of manual tasks.

The AI Copilot Experience

Based on that, what data and what data systems can we bring in, how can we model and mold it, and then now provide a copilot experience or a reporting experience for a user that doesn’t then have to extract data into a spreadsheet or copy and paste data into a screen. How do we get them to actually making decisions and taking actions without having to worry about that back end process?

I’m a nurse what does AI mean to me? What am I doing differently than what I do today?

I think some of the use cases that I’ve seen in senior living or long term care is a nurse is having to wade through tons of different documents as they’re looking to admit a patient, let’s say or understand, what’s going on with this patient.

What have they done for the past week or what’s their care plan? What’s their meal plan? And then have to actually put those things together for the resident or the patient. And what the copilot can do can be the means of putting together that care meal plan for them much more quickly, putting together a rough draft.

It can look across the different systems, the chart notes, and typically you’re getting PDFs, whether it’s from the emergency care, like the emergency unit that they had to go see 2 weeks ago, or, hey, they’re getting care from outside our 4 walls, what happened without the nurse having to go look through all those different charts and summarize it for themselves. So how are we bringing this into their own workflow? Another thing we’re seeing is scheduling. if you’re scheduling the labor for who is working with what resident for the week. It’s also another thing a copilot can help with.

History of admissions

Amber Bardon: So just going back to your previous example on the admissions or just sorting through the data of what’s going on with the resident. I’m just imagining that what Skypoint would do and correct me if I’m wrong, is they would have the ability to go to this AI tool and say to it, “Give me a history on Mrs Smith and give me the history of what led to admission” and then it would pull the data from all these different sources and essentially give you that little paragraph of summary of data.

And then the same thing would be, if they want to look back, maybe all our clients do care meetings, every morning or shift change meetings, they could say give me a summary of everything that’s gone on in the past 24 hours without having to go pull the clinical notes and look at the chart and pull the assessment.

Gregory Petrossian: That’s exactly what I’m describing. And I think in addition to that, think about the communication that needs to happen with the family and how can you more easily put that plan or what’s been happening and then communicate to the family more easily.

“Hey, help me write an email to the resident’s family”, or “Hey, a resident’s family asking about a policy or procedure, or what’s gone on. Can the copilot just write that quick response for me?” And so let’s increase that touch that we can have and that communication across all aspects of the care.

So how does this work?

How does that piece work? how do you build this on the backend for a client? What does a client need to have in place to be able to take advantage of this technology?

So the way that end consumer interacts with the experience can be the choice of the operator, whether it’s working with ChatGPT or what Microsoft is releasing called M365 Copilot.

We can also build what’s called a custom copilot with Microsoft’s low code technology called Copilot Studio. And then finally, we have built our own product called our private copilot, where we can actually white label and deploy a web application with a custom URL for the operator, like, and it looks like one of their own applications in their environment.

You have different options for that experience, and there’s different licensing models that everybody has, so we just help bring the right experience and model to the operator.

Bringing the infrastructure together

Amber Bardon: So to go to the 2nd, part of the question, tell me what this looks like from an infrastructure perspective. Let’s say a client has PCC and Sage Intacct and OnShift and Paycom and Sherpa and eChoice Menu. How is this all working together? Do you have to have agreements with each vendor individually? Just walk me through with this. Like, how do you build this?

Gregory Petrossian:  What we provide and what our platform truly is all the back end infrastructure where data can be brought in and unified and if a customer has these relationships with all their vendors, they need to ask them for data access, and some vendors are easier, and some vendors are more difficult to get that data access, and sometimes you have to pay them money for that data access, and every vendor provides data in a little different way.

And sometimes they have different options for getting to that data. And that’s where we help as the advocate of the customer. We say, we walk through and talk with all their vendors and ask these questions. Typically, our customers or operators don’t have the technical depth to even know what to ask for.


So we help them in that regard. And we’re building integrations with Many of the data systems that they’re familiar with currently and have worked with . As soon as you get that access, whether it’s through an API, some SFTP or you’re getting just CSV files in some way. We work to automate that data ingestion, bringing that data into that unified back end and what we have is a pre built architecture that plugs into that copilot experience all the different ones I was mentioning as well as we’re building and massaging that data that could be used for a copilot, or it could be used for reporting the way I like to explain it is you want to build your data warehouse or your data lake house these data structures in a way that can be used in different modalities.

Whether it’s for generative AI or whether it’s for reporting. Build it once and use it across the different ways, whether you’re trying to explore data with reports or whether you need a specific answer, like you were saying with your Google search, or you need help with a specific workflow. So we bring all that and put it in a box and with clicks of a button as a customer purchases it.

Maintenance & Support

We deployed that infrastructure specifically into their four walls. That’s what makes us different is we are an enabler for data, and we help accelerate the path to doing that because a lot of my background and experience has been building these things from scratch. There’s a lot of work you have to do to be able to even get to the point of being able to bring data in the door and when we had all these different customers on one off builds, it was hard for us to manage and maintain that.

So a lot of the value we provide is consistency in how it’s built and that infrastructure. So we can more cheaply provide support and maintenance and break fix. We have a fractional team that can support the end to end data process that gets that data to the large language model or to the report. And that’s what we’re making cheaper and easier for operators to be able to compete with the larger operators that have the money to build these things from scratch or have data teams, we want to be the easy button for them to be able to get going on these type of efforts.

How many senior living communities are you currently working with?

Gregory Petrossian: Right now we’ve got four or five operators we have been working with for the past year and a half, and we’ve got another five to ten operators that are in our pipeline. So I would say it’s still early days for us and I think it’s still very early days for generative AI, and we’re really working with the operators that are pioneering and leaning in. And I think that’s the difference is, sometimes I’m talking to operators, they’re like, Hey, give us the use cases and the things that this works really well for. This has only been out for, 12 to 18 months. And so we’re working with you on tackling the use cases you are seeing and the pain points you have. And so those operators that are wanting to be more on the bleeding edge are the customers we’re seeing.

AI use cases for the residents

Amber Bardon: Do you ever see. Potential in the future use case for the residents themselves to have access to this type of tool.

Gregory Petrossian: Absolutely. What I’m telling operators is you want to solve this first internally on your internal data and with your internal teams before you start deploying and putting it out there in the public. An example I like to pick on is, I was seeing these memes of things like some Chevrolet dealership that had taken their customer service chat bot on their public website and was calling some open AI API and people were using that chat bot as just like a free chat, GPT pro subscription and asking it things well beyond, “Hey what is the price of this model or what is the price of this car”. Or “hey, when can I buy it?” They’re telling you to do its homework, right? Because they didn’t put the right guardrails and the right context for it to be specific to what it needed for that dealership.

Preventing Data Hallucinations

And so that’s what we’re doing first is there’s a lot of training and reconciliation and fine tuning we have to do to make sure that the AI copilot is reliably responding and what we have done with our product is effectively turn off hallucinations. And so you’ll get more. “Hey, I don’t have that answer” then you would with chat That’s going to make up an answer. And we go through a testing phase, very… I would say an iterative sprints or iterative cycles where we want to take a small piece of their data and use case, solve that first and then start building on that rather than boiling the whole ocean all at once.

And so it is an iterative process to get to the endpoint, and you’re never really done with your data. Whether you’re doing data warehousing analytics, business intelligence, AI, you’re always building something and iterating on it, and it’s never really done because your business is always pivoting and changing.

Your data systems are always changing. How can you bite off enough chunks that you can actually chew and get through and see value on the other end? Because if you never see business value on the other end is what really kills these projects. So we’re very meticulous and what we’re going to take on with our customers and operators because we need to see true business wins.

What’s Next for AI Adoption?

Amber Bardon: I think you’re absolutely right that this is definitely very new and, it’s really exciting to think about this being the future. Essentially that we’re taking the technology that we’re all just getting now on our phones and applying this to the industry at whole and it’s been such a huge gap.

Data analysis has been such a huge gap in the industry for years and years. And there, there hasn’t really been a great solution. And I almost feel like this is leapfrogging some of the other things like just BI tools, Power BI and things like that.

This is all very new and it is the future, but what do you see beyond this even? So after this becomes more commonplace and we start to see more and more operators adopting this, what do you think is next?

Gregory Petrossian: What I think is the future of AI and AI copilots is we’re going to see organizations having these AI copilots to be specialized within their business. You’re going to have a copilot that really knows your HR data and your HR processes. It knows your sales and marketing department really well, and it’s helping them. And so you’ll see business domains having these specialized copilots helping in their day to day, their analysis, their forecasting and that’ll become way more commonplace as reporting business intelligence has been already.

Automated AI Agents

What we’re going to see as well as the next stage in what copilots can do is automated agents actually doing these tasks, having a level of autonomy of, ” can you work with this resident’s family on any of their questions” and it can go and scan. Oh, I see this email that came in and I’m going to respond based on the resident’s information, the context I have, and it’s going to be able to do that more automatically.

And so I think, again, that’s what’s really scaring people is that it can do more and more task oriented things on its own. But I think we’re pretty far away from that. And it’s the tasks. That’s really the grunt work that we don’t really want to do or should be doing. And we really want to lean more on the human touch and the human relationship in senior living, actually caring for the resident. Rather than the administrative back end back office test that we are grown to do.

I think it’s an exciting time for operators to lean in to their data. I think it’s an exciting time just for health care in general, where we can finally realize what we’ve been saying for so long of being able to provide better care for cheaper price and to really be, value based in how we operate.

So, excited to see what the next 5 to 10 years holds for senior living at large.

Share This:

More Resources


Your Unified Data, Analytics & AI Partner

Industry leaders choose Skypoint as their comprehensive and compliant Modern Data Stack Platform. A Certified Microsoft Solutions Partner, Skypoint Cloud turns siloed data into connected experiences.