How AI is Reshaping Home Care at Three Levels - with Stephen Tweed

Ken Accardi: Welcome to the next installment of the Home Care Heroes podcast. We have a young man here named Stephen Tweed, who has been in the industry of healthcare at home for 40 years. For 25 of those years, he's led a team called Leading Home Care, and pretty much everybody in the industry knows Stephen. He put in motion in 2013 something called the Home Care CEO Forum, and he is a top industry researcher.

I've actually learned a lot from Stephen over the years and used a lot of his metric systems. For example, we took the 12-point system that Stephen put in place for the best and most important metrics a home care agency should measure - that came out of one of the CEO Forum groups - and we made sure that we could answer all those questions with metrics in our own software at Ankota. So we're very privileged to have Stephen here.

Now, this young man who's been in the industry for a while - we're actually going to talk about a very relevant and timely topic, which is artificial intelligence. Stephen has stepped out of doing the direct home care forums, but he's doubled down on research. He says he's not retiring, he's rewiring - figuring out new ways of doing things. He has really jumped in with both feet into artificial intelligence, and that's what we're going to talk about today.

So first of all, welcome Stephen. Thanks for being on Home Care Heroes.

Stephen Tweed: Well, thank you so much. It's always a pleasure to be with you and your audience and talk about things that are mutually interesting. And as we've agreed over the years, there's lots of fascinating stuff going on in home care.

Ken Accardi: One thing that you mentioned before we started is that you've really found that artificial intelligence can help home care agencies on three levels. Can you talk a little bit about that?

Stephen Tweed: Sure. As I've been studying and trying to get a grasp on this, I came across a wonderful book by a university professor from the University of California at Berkeley. His name is Sangeet Paul Chaudhary, and his book is called Reshuffle. He was talking about how AI and technology are reshuffling major corporations. I was intrigued because what I learned is you don't need to be a guru or a geek to understand the principles of how you apply AI to a business organization.

As a 40-year business strategist in healthcare at home, I've always been looking for new insights into how leaders develop and implement strategies to grow their business, to gain competitive advantage, to differentiate themselves in the market, and to fulfill their vision of what they want to become.

As I was reading Chaudhary's book, he made it very clear that AI really fits into an organization at three different levels.

The first level is individual tasks. In home care, there are lots of people doing individual tasks - whether it's recruiting caregivers, retaining caregivers, scheduling clients and caregivers, doing sales and marketing activities, or even billing. There are individual workers performing individual tasks, and AI can improve the speed, reduce time, and save costs for those individuals.

One of the real insights that came out of all my studying is a clear understanding that, at least in the foreseeable future, AI is not replacing people. It is making them more effective, more efficient, reducing time, saving money. The people will still need to be there. Now, somebody said to me, "AI won't replace people, but people using AI will replace other people." So I think one of the lessons for your listeners is to be thinking about how to use AI in your company and how to help your team members be more efficient and more effective.

The second level is organizational systems. Coming out of the Top 5% CEO Mastermind Group over the last decade is the lesson that leaders of large home care organizations are systems thinkers. They think in the context of: if there's a task that we do every day, we need to have a system. A system is a combination of process and people - a step-by-step process that we do over and over again, performed by people who have the skills and discipline to apply that process consistently.

Within a small home care company, a scheduler has a process where they're doing scheduling the same way over time. But when you grow to two locations, four locations, or 25 locations, you need to be doing that same system the same way across all locations.

Years ago, I identified what I call the five phases of flow of a home care company. Every client goes through these five phases:

1. The attraction phase - sales, marketing, SEO, social media, and all the things we do to attract new clients.
2. The conversion phase - where we convert callers to clients, turning leads and inquiries into clients receiving care.
3. The staffing phase - caregiver recruiting, retention, onboarding, and training.
4. The caregiving phase - scheduling, care coordination, time and attendance reconciliation, EVV, and all the things you do to make sure people are on the job doing what they're supposed to do.
5. The collection phase - billing, collections, financial reporting, and everything that happens at the back end.

We've been working with those five phases of flow for years, and when we started thinking about systems, we took each phase and broke them down. We identified 12 specific systems that a high-performing home care company needs to have in place: a sales system, a marketing system, a recruiting system, a retention system, a care coordination system, a scheduling system, a billing system, a collection system, a reporting system, and more. Not every company has all 12 systems in place all the time, but it's what we work toward.

Then we look at AI in the context of how this new technology helps us improve those systems. Remember the old continuous improvement principles? How do we use AI to continuously improve our systems inside the company to increase performance and reliability?

The third level is what Chaudhary calls the competitive ecosystem. How will AI help organizations restructure in a way that makes them more competitive in the marketplace? How do they restructure the way they do business to create strategic market advantage? That's the exciting opportunity down the road. It's going to be a while before companies in home care get to that level, but AI has the potential to totally reshape the home care competitive ecosystem.

So tasks, organizational systems, competitive ecosystem - those are the three levels where AI is going to evolve and develop in home care. Lots of opportunities, and a long way to go.

Ken Accardi: That's a great rundown. So just to recap: we're looking at three levels - tasks, systems, and then how we could build competitive advantage. Let's start with the tasks. If we'd spoken about AI even two or three years ago, the first answer you might have heard was, "We are a personal relationship, hands-on business where we send caregivers to homes to take care of older folks who don't really think about technology." I think that's totally changed.

So let's stick on the task level for a minute. What are a few of the tasks where you've seen agencies getting real productivity out of using AI to streamline things? What's the low-hanging fruit for folks who are ready to dip their toe in?

Stephen Tweed: The three big areas are sales and marketing - the whole client attraction area. Second is caregiver recruiting and retention. And third is scheduling and care coordination.

Let's take a recruiter in a home care company. What we know is that most home care companies are spending a lot of their time recruiting online. The data from Home Care Pulse and Activated Insights show over the last 15 years that the number one online recruiting source is Indeed.com. And yet the recruiting source with the highest turnover is also Indeed.com. The recruiting source with the lowest turnover and the highest quality applicants is employee referral systems. But many companies don't put a lot of emphasis on their employee referral system because it's easier to get lots of applicants online.

So we step back and look at how to use AI to improve online recruiting, how to improve employee referral programs, and how to improve face-to-face recruiting in the community or on campus. An individual recruiter can start putting some of those questions into ChatGPT or Perplexity and begin to get ideas and information.

Part of the understanding that people in home care need to grasp is that AI looks at lots of data and information, quickly analyzes and synthesizes it, and gives you answers more quickly than a human being can do it wading through the data.

I'll give you a classic example. I was at a symposium put on by the University of Pittsburgh Medical Center, and a group of physicians were presenting how they're using AI in medicine. They said only 34% of physician interactions where the physician prescribes a medication or therapy are based on actual medical knowledge. So 66% is based on something else. A physician has three sources of information: medical journals, medical textbooks, and classroom learning. But with all the material out there, my family physician doesn't have the time or capacity to read and remember everything. So my doctor is recommending what worked last time, or what a pharmaceutical rep said works, or what a colleague is trying. So much of it is trial and error, and only a small percentage is based on actual medical knowledge.

If we apply that to home care - what about scheduling and matching clients and caregivers? How much of what we do is based on actual information and knowledge about the client, the caregiver, distance, traffic, and all the factors that affect whether that schedule works? And how much is intuitive? We see the same thing in recruiting, where the recruiter often hires based on intuition rather than good data, because getting the data takes too much time. They hire the people they like rather than the people who have a demonstrated ability to be good caregivers and to be reliable.

Ken Accardi: That's great. From my perspective, I use a lot of AI for a lot of things. ChatGPT is probably the most accessible and the main brand name that most people know, and it's probably the best conversationalist. You can click the microphone and talk into it, take a break, click the microphone again - it's very conversational, and you can run it on your phone.

For example, Claude AI, which is what I use mostly for writing software, requires a bit more typing, but Claude is a much better writer than ChatGPT. So I'll ask ChatGPT to strategize things, then ask Claude to do the final writing, and I might even proofread it with Gemini. When it's going on the web, Gemini has the best image generator, called Nano Banana of all names. You can say, "I wrote this piece for our marketing, could you read it and suggest two images?" and it'll create beautiful images.

If you haven't tried anything yet, the first two things I'd recommend: start with a job description. Take your existing one and ask the AI to make it fantastic. Maybe say something like, "I heard Nick Bonitatibus talking about making the job description be like an irresistible offer for a caregiver to want to come to the company. How can I do that?" The AI will write it in an amazing way.

Another one: if you have a new state regulation and need to incorporate it into your policies and procedures, you can give your policies and procedures to ChatGPT, give it a link to the new regulation, and say, "How would you recommend that I incorporate these changes?" These are everyday tasks where you might not think AI could help, but it can do a fantastic job.

AI has gotten so much smarter in the last year or even six months. It used to be that you'd give it a spreadsheet and it couldn't understand it, but now you can give it a pretty big spreadsheet and it'll help you understand things like where to focus on improving collections.

Stephen Tweed: As we move forward, it's really going to be about how we automate those processes so that on an ongoing basis, the AI platform is looking at information in real time, constantly updating, learning, and growing. Six months ago, AI made lots of errors and we didn't trust it. Now it's getting better because it's learning from the data we put in.

Let me think about a simple system any home care company could start with. If we start at the beginning with sales, one of the things I've been doing with a number of companies is talking about selling home care to your ideal client.

How do you find your ideal client? In any home care company, the ideal client is based on the referral partner that brings you the largest number of new clients with the largest number of hours per week and the longest length of stay. From an operating effectiveness point of view, it's much more effective to have one client with high hours per week who stays with you for a year, two years, or three years, than to be constantly churning clients with three hours a day four days a week, trying to match caregivers through all those scheduling challenges.

So we look at a system that regularly gathers admission and discharge information from our clients and analyzes who has the highest hours per week and the longest length of stay, and how did they come to us? Who are the referral partners? Did they come from online? From a hospital discharge planner? From a bank trust officer or an estate attorney?

By analyzing that data quickly, we can send information to the sales rep that says: the highest potential for how you spend your day is to go out and identify what we call the trusted advisors - the bank trust officers, geriatric care managers, and estate attorneys - because our data show that those are the referral partners that bring the highest hours per week and the longest length of stay. We can drill down not only to the category of referral partner, but to the individual organizations and facilities where those people work.

That's an example of how we take a task and build a whole system for selling. Looking at the same thing for scheduling - the most recent data from our research suggests that the number one application for AI and systems will be client-caregiver matching and scheduling. How do we build an operating platform that uses AI to take a client with particular wants and needs and a caregiver with particular wants and needs and match them up, making sure that client and caregiver are together as often as possible?

Our earlier research identified that what clients want from a home care company is, number one, quality care - basically that the caregiver is helping them with quality of life and independence in the home. Second, they want reliability - a caregiver that shows up every day as scheduled, on time. And third, they want continuity of care - the same caregiver every time.

How do we use AI to do that match so the client and caregiver are together every shift? Obviously, it's easier if you have a large number of clients with high hours per week and a large number of caregivers with high hours per week.

We've talked about the rule of threes - three types of caregivers in home care:

1. The full-time professional caregiver - somebody committed to being a caregiver who loves their work and wants to do it 40, 50, or 60 hours a week. They're going to do this work regardless. The only question is which agency they work for.
2. The halftime professional caregiver - committed professionals who want to work 20 or 25 hours a week. Many are in a second career: retired school teachers, retired police officers, retired paramedics, or stay-at-home parents who want to put their kids on the school bus at 9 and pick them up at 2:30, but will work a four-hour shift in between.
3. The part-time intermittent caregiver - they work when they want to, when they're available. They're the ones that disrupt the scheduling process because you can't always count on them. They're more likely to call off. When you call them and their availability shows they can work Thursday from 10 to 2, they say, "No, my kid's got a thing at school."

Using AI to really fine-tune that process is powerful. And then there's the distinction between generative AI and agentic AI, where the AI actually becomes an agent. What we see coming down the road in scheduling is where the software sends a message to the caregiver saying, "You said you wanted to work Thursdays, and we have a shift. What's your availability?" It sends that out to 10 caregivers, three come back saying they're available, and now the human can look at the match information and make a decision: I'm going to put this client with this caregiver and communicate with them both.

We're already seeing some companies doing this in their organizational systems. We have a long way to go, and you can't do all 12 systems at once. You have to focus - whether it's your scheduling system, your sales system, or your recruiting system - and figure out how to use technology and AI to improve that system.

Ken Accardi: I love it. I actually listened to a fantastic podcast in our industry - the Miriam Allred podcast, now called the Home Care Strategy Lab. She recently interviewed the CEO of a new company called Phoebe. Phoebe is looking to solve some of those scheduling problems. What they drill into is that when you get a caregiver cancellation 90 minutes before a shift, you might have to call 11 to 14 people to try to fill it. Phoebe's AI can first try texting people and then actually make phone calls where the caregiver is talking to AI - not only trying to fill that shift, but also taking that caregiver's wants and needs to the next level. They're asking, "Are you okay with dogs?" "Well, I'm okay with little dogs, not bigger dogs." They're using AI to store data about a caregiver that was never captured before, so the next time they're filling a shift, that information helps. That company raised $9 million and they're off to the races.

There are a lot of things that weren't everyday software even five years ago. Micro caregiver rewards platforms are new. Phoebe with scheduling is new. There are dedicated recruiting platforms. There's even a company that has a live chat component for your website, so when somebody is looking for home care at 10 o'clock at night after getting their kids to bed, they can actually speak to a live person.

The work you're doing, Stephen, drilling into these business systems and looking at the data to find opportunities - finding your ideal customer profile, changing processes to spend the right amount of time on top referral sources - that's a great use of AI. Or taking something like a scheduler who was going to work on improvements or recruiting, but suddenly has to spend an hour and a half on the phone trying to fill a shift - having AI help with that. These are really great solutions coming down the pike.

I don't want to go too crazy with this, but there was this guy named Elon Musk who said we could combine AI and battery technology with automotive, and now we have self-driving Teslas. His next thing is figuring out how to put data centers in outer space - he just merged SpaceX and xAI to start using satellites powered by solar energy for AI data centers. We're not going to come up with those things in home care, but if we think about converging things that weren't there in the past, we're going to come up with some great new innovations.

One thing my company is working on is a combination of an AI companion and emergency response. Basically, using the ability for an older person to speak to the AI and have conversations with it, but also have an emergency response button where the AI will respond first. While it's doing this, it's listening in - but it's not spying on the person. We've worked really hard to respect privacy. If grandma wants to have a boyfriend or a shot of Bailey's before bed, that's her personal business. But if she's talking about not feeling well, or somebody called asking about her bank account, those are things we would escalate to the home care agency and the family.

We're trying to think in different ways. I think the best thing we can do in home care now is think out of the box. We're the most important industry for the next 20 years with the population growth we have. Figuring out how to do things with fewer caregivers - there's no worry about job security in home care. The challenge is how to be as productive as possible, because there are a lot of people we can provide care and help to.

Stephen, we started with one question and you covered a lot of amazing ground - the different types of caregivers, the main processes and systems. We're rooting for you to find great solutions and point us in the right direction. Do you have one last story to put a capstone on this episode?

Stephen Tweed: You were talking about the different technology and about listening in to the client at home and sorting out what the technology hears to protect their privacy. That's one of the things I've observed over the last decade. I've probably had a hundred conversations with founders of new companies looking at using technology to transform home care - various sensors, cameras, motion detectors, audio recording. Very few, if any, have gotten significant traction in the industry. The big reason is that clients resist the technology because they want to protect their privacy.

I remember years ago visiting with my grandmother in an assisted living facility. I was speaking at a conference of senior living communities, and I said, "Grandma, I'm speaking at this conference next week. What should I tell them that's most important from your perspective as a resident?" She thought for a moment and said two things:

"Don't shout. I'm not deaf." Her experience was that people would come into her room and speak loudly because they assumed she couldn't hear.

"Respect my privacy. This is my home." Knock when you enter. Be respectful.

That was probably 25 years ago, and it stuck with me. It resonates in the context of how we apply AI technology: we need to understand the needs of the client, the patient, the resident. We don't need to speak loud because they're not deaf. And we need to be sensitive to the privacy issue.

For home care owners and leaders and for people supporting the industry - when looking at how to apply AI, I think the question is: how do we get the technology to perform tasks, operate systems, or create competitive advantage in a way that respects the needs and wants of the client and their privacy? Those seem like two simple things, but they're some of the biggest challenges as we go forward.

Leaders need to be thinking differently about how to use this new technology, recognizing that it's changing every day. AI is constantly learning, so the answer you get next week may be different from the answer you get this week to the same question. We need to be learning, growing, and applying AI in new and different ways to transform the home care system and the competitive ecosystem.

Folks like you in the technology side of home care have lots of opportunity to make a difference for your clients and for the individuals within those organizations - helping them serve their clients, their caregivers, and their community.

Ken Accardi: Thank you. Stephen, you've always been so gracious sharing your knowledge, for at least the 10 years I've known you. If folks want to get in touch with you and your research, how would they do that?

Stephen Tweed: My company is called Leading Home Care, a Tweed Jeffries company. My wife and business partner, Elizabeth Jeffries, is the partner in our parent company, Tweed Jeffries LLC. Leading Home Care is our practice that focuses on the home care space.

The website is leadinghomecare.com. You can find me on LinkedIn. You can subscribe to my weekly newsletter called Stephen Tweed's Thursday Thoughts that comes out every Thursday - just go to leadinghomecare.com and put in your email. That's all we need; we're not going to spam you or call you or bug you.

If folks want to have a conversation, my email is stephen@leadinghomecare.com - Stephen with a PH. I'm always open to hearing from folks and having a conversation. If I can help, I'd love to, and if not, I may be able to find someone who can.

Ken Accardi: Perfect. Thank you so much, Stephen Tweed, for being here on Home Care Heroes and Day Service Stars. Have a great day.

Stephen Tweed: Thank you. Great day. Bye-bye.