

5 min read
A plain look at the work these agents take off your desk, the work they hand back, and how to tell the two apart before you buy.

Co-Founder & CEO
AI agents for home health agencies run the front end of intake and hiring: answering instantly, qualifying, and booking the next step. In 2024, spending at freestanding home health agencies grew 10.2% to $169.4 billion, so the volume arriving at that front door keeps rising.
The Two Jobs Worth Automating First
An AI agent earns its place in a home health agency on two jobs: intake and hiring. Both run at high volume, and both are repetitive at the start and judgment-heavy at the end. Both are also lost when nobody answers fast enough. Everything else an agent could do is a distant third.
Start with how much is moving through that front door. The Centers for Medicare and Medicaid Services, in its National Health Expenditures 2024 Highlights, put spending on services from freestanding home health care agencies at $169.4 billion for 2024, a 10.2 percent rise on the back of 11.1 percent the year before. Two consecutive years of double-digit growth means more referrals, more family calls and more applicants, all arriving at the same intake coordinator who was already at capacity.
That growth isn't evenly useful, though, because an agency can only convert what it actually answers. Spending and referral volume are rising faster than most agencies have added intake and recruiting staff, which makes response capacity, not demand, the thing that decides how much of the growth you capture.
An agent is a capacity tool before it's anything else.
It helps to be precise about the word "agent," because the market isn't. A scripted chatbot answers questions. An agent carries a task to completion: it holds a conversation, applies your rules to what it hears, then acts in another system. It also knows when to stop and fetch a person. That last ability is the one operators underrate, and it's the one that decides whether the tool survives a month of real referrals.
The two jobs also feed each other in a way vendors rarely mention. You can't accept a new client without the staff to serve them, and you can't keep staff busy without a steady flow of clients. An agency that automates only intake fills a pipeline it can't service. One that automates only hiring staffs up for referrals it never answered. If you are weighing where to start, our breakdown of how home health agencies turn referrals into admissions walks the intake side end to end.
What Happens to an Inquiry That Arrives at 7pm?
Usually nothing, and that's the whole problem. A discharge planner sends a referral at 7pm. An adult daughter fills out your contact form on a Sunday. Both sit until morning, and by then at least one of them has called the agency down the road. That isn't a staffing failure. It's a coverage gap, and it's the single clearest use case for an AI agent.
What an intake agent actually does in that moment is narrower than the marketing suggests, and more useful. It answers immediately. The questions that follow are the ordinary ones: what service is needed, where the client lives, who's paying, when care should start. That gets checked against your service area and your acceptance criteria, and then the agent either books a call with your intake nurse or routes an urgent referral to whoever is on call.
It isn't deciding whether to accept the client. It's making sure that decision reaches a human with the information already gathered.
Speed matters because the people on the other end expect it. Pew Research Center's Mobile Technology fact sheet, published in November 2025 from a survey of 5,022 US adults, found that 78% of adults aged 65 and older own a smartphone, alongside 90% of adults aged 50 to 64. The family member coordinating care for a parent is reachable on a device in their hand, right now, and they know it. A next-business-day callback reads as disinterest rather than diligence.
Qualification is the part that separates an agent from an answering service. An answering service takes a message. An agent asks the four or five questions your intake coordinator would have asked anyway, so the callback starts from a real picture instead of a name and a number. Alita's intake agents run that qualification against criteria you set, and the whole conversation lands in the operator dashboard rather than a voicemail box.
How AI Agents for Home Health Agencies Handle Hiring
Hiring is the job where the volume is genuinely relentless, and where an agent pays for itself fastest. Federal labor projections covering the decade from 2025 to 2035 rank home health and personal care aides among the fastest-growing occupations in the country, with annual openings counted in the hundreds of thousands. That's the market you're recruiting into, and it doesn't let up.
The figures behind that are worth knowing precisely. The US Bureau of Labor Statistics, in its Occupational Outlook Handbook entry for home health and personal care aides, projects employment will grow 18 percent from 2025 to 2035, with about 760,500 openings each year across the decade.
Openings on that scale are driven as much by churn as by growth, which changes what the work feels like. PHI, working from a slightly earlier 2024 to 2034 baseline, projects 9.7 million total direct care job openings in its Direct Care Workers in the United States: Key Facts 2025, and puts median annual earnings for direct care workers at just under $26,000. Turnover is easing but still high: Activated Insights reported in its 2025 Benchmarking Report that home-based care turnover fell to 75%, the lowest level it has recorded in five years.
Set median pay just under $26,000 against 760,500 annual openings and you get an uncomfortable but clarifying conclusion. No single agency can out-pay the shortage, because the constraint is sector-wide compensation rather than any one employer's budget. What you can control is narrower: how fast you reply, and whether anyone follows up when a good applicant goes quiet. Those are process levers, and they're exactly what an agent operates.
Candidates say much the same thing. In its 2023 Ghosting in Hiring report, the job board Indeed found that 62% of job seekers plan to ghost employers in future searches, up from 56% in 2022. Read it as a vendor survey of its own users rather than independent research, but the direction matches what recruiters describe, and the fixes employers named were communication and speed. A hiring agent answers an application the minute it arrives, screens for licensure, availability and service area, and books the interview while the candidate is still interested. We covered the wider staffing picture in our look at smarter hiring for home health agencies.
See Alita in action
Book a quick demo and watch every inquiry turn into a booked tour, consultation, or interview.
Which Channels Does Your Agency Actually Need to Cover?
More than your website. A referral arrives by phone. A family messages your Facebook page. A caregiver replies to an Indeed post at 11pm. Each of those is a door. A web-chat-only tool guards exactly one of them, and the rest stay unattended.
Texting deserves particular attention because it is where care conversations actually happen. SimpleTexting's 2025 Texting and SMS Marketing Statistics report, based on a December 2024 survey of 1,000 US consumers, found that 82% of consumers check their texts within five minutes and 71% want the ability to text a business back. Treat that as a directional consumer benchmark rather than a clinical finding, since it is a vendor survey, but the direction is hard to argue with.
The practical list for a home health agency looks like this. Website chat for the families researching you. SMS for the follow-up that actually gets read. Phone coverage, because referral sources and older clients still call, which is what voice agents are for. Social messaging on Facebook and Instagram, where adult children already spend their time. And link agents attached to job postings and ads, so an applicant who taps an Indeed listing lands in a real conversation instead of a form.
One coordination point matters more than the channel list itself. If each channel runs on its own tool, you have rebuilt the original problem with extra logins: the family who texted on Friday and called on Monday looks like two strangers, and the applicant who messaged Instagram before applying on Indeed gets screened twice. Coverage is only worth buying when the conversations land in one place with one history.
Link agents deserve a closer look. Job boards send traffic to a posting, the posting sends it to an application, and the application goes into a queue. Every step loses people. Putting an agent at the click means the screening conversation happens at the moment of highest interest, which is the only moment you reliably get. Our home health agency overview lays out how the channels connect to one pipeline.
What AI Agents for Home Health Agencies Should Not Do
An agent should never make the clinical call. It doesn't decide whether your agency can safely take a client with complex wound care needs. It doesn't verify a license, authorize benefits, or complete an OASIS assessment. Anyone selling you those capabilities is describing a liability, not a product.
The boundary is cleaner than it sounds. It maps to one question: is this step a judgment or a gather? Gathering is agent work. Judging is human work, every time.
Step in the process | Agent or human | Why the line sits there |
|---|---|---|
Answering a 7pm referral or application | Agent | Availability, not judgment |
Asking service, location, payer and start date | Agent | A fixed script your staff already runs |
Deciding whether to accept a client | Human | Clinical capacity and safety judgment |
Verifying a license or certification | Human or primary source | Compliance record, not a conversation |
Booking the interview or intake call | Agent | Calendar work with clear rules |
Assessing a candidate's fit and judgment | Human | The part hiring actually turns on |
Two softer limits belong on the same list. An agent shouldn't improvise about what your agency covers, so it needs your real service area, payer mix and acceptance criteria loaded before it takes a single call. It also shouldn't hide that it's software. Families coordinating care for a parent are already anxious, and discovering halfway through that they've been talking to a bot costs more trust than the speed gained.
Burnout is the reason this division matters rather than a nice-to-have. The Centers for Disease Control and Prevention, through NIOSH, reported in its 2023 Vital Signs analysis of the Quality of Worklife Survey that 45.6% of health workers felt burnout often or very often in 2022, up sharply from 2018, with the share saying they were very likely to seek a new job rising from 11.1% to 16.5%. Taking repetitive work off your intake and recruiting staff is a retention strategy, not just an efficiency one.
How Do You Tell a Real Agent From a Chatbot?
Ask it to do something. A chatbot answers questions from a script, then hands off. An agent qualifies against your criteria and writes to a real calendar. It routes by urgency, and it follows up on its own when someone goes quiet. The difference is whether the thing can complete a task or only discuss one.
Five questions sort the market quickly:
Does it cover phone, SMS, social and job links, or only your website?
Can it write to a real calendar, or does it just collect a preferred time and leave someone on your team to do the actual booking later?
Intake and hiring both, or two systems?
What happens the moment it hits something it can't handle: does it page a named person, drop the conversation into a queue nobody watches, or keep improvising?
Where do the conversations land, and can you read them without exporting a file?
Watch for a specific trap in demos. Plenty of tools perform beautifully on the happy path, then collapse on the messy referral or the applicant who's licensed in a neighboring state. Ask for those. Escalation behavior tells you more about a product than any scripted walkthrough will.
Measurement belongs in the same conversation. Four numbers tell you whether an agent is working at ninety days: share of inquiries answered within five minutes, share qualified before a human touches them, interviews booked per hundred applicants, and how often the agent escalated when it should have. Any vendor that can't show you those numbers in its own dashboard is asking you to take improvement on faith.
Alita was built for both jobs on one set of agents, across every channel above, which is a deliberate answer to the two-vendor problem. Whether you buy that or something else, the standard is the same: the agent covers the hours and the volume, and your clinicians and recruiters keep the work that needs a person. For a narrower view of the intake side, we went deep on what a home care referral AI agent can actually handle.
Summary
AI agents do their best work for home health agencies on two jobs, intake and hiring, and both come down to answering fast and qualifying well. The volume is real: CMS put 2024 spending at freestanding home health agencies at $169.4 billion after 10.2 percent growth, and the Bureau of Labor Statistics projects about 760,500 aide openings a year through 2035. What an agent contributes is coverage across phone, text, social and job links, plus the screening questions your staff would ask anyway. What it should never contribute is a clinical decision, a licensure verification or an assessment. Judge any vendor on whether it completes tasks and escalates cleanly, not on how well it talks.
More from Alita
See all blog posts
4 min read
Why AI Chat Agents Outperform Offshore Human Agents in Home Care

2 min read
Alita Accepted into AgeTech Collaborative™ from AARP Accelerator Program

5 min read
15 High-Impact Places to Display Positive Reviews for Home Care Agencies

5 min read
The Essential Guide to Google Reviews for Home Care Providers

8 min read
The Essential Guide to Ranking on Google: For Home Care Providers

4 min read
Alita Welcomes Longevity Market Pioneer Mary Furlong to Advisory Board

