Blog
Why frontline care workers are abandoning AI
Most organisations roll out AI and call it a win when staff log in. But if that’s how you’re measuring success, you might be missing what’s really happening on the ground.
Here’s a pattern that plays out again and again in aged care AI rollouts: administration and finance teams adopt quickly. Frontline care workers don’t. And the gap between those two groups, if left unaddressed, is often where the value of an AI investment quietly disappears.
That gap doesn’t happen by accident. And understanding why it exists is the first step to closing it.
Learn how to implement AI for your team in our recent webinar, Implementing AI: Moving your team from ideas to impact:
Not everyone experiences AI the same way
When an AI program lands in your organisation, it doesn’t land equally.
Your head office and finance teams have built-in advantages. They sit near each other, can ask colleagues for help and tend to be more comfortable with new digital tools. They adopt quickly.
Your frontline care workers, the ones travelling between client homes or moving from room to room in a residential facility, don’t have those same supports. They’re often working alone, pressed for time and dealing with technology that feels unfamiliar.
“The change across your organisations based on what we call digital readiness varies quite a lot,” says Ali Kaabi, Director of Client Solutions at Rohling. “It is very normal for your back office and head office individuals to be the early adopters. That’s not necessarily the case when it comes to frontline care workers.”
Ali’s point is clear: if your adoption metrics are showing green early on, it’s worth asking: which cohort is actually driving those numbers?
The middle management squeeze
Before you get to frontline workers, there’s another group that’s often overlooked in AI rollouts: your facility managers, care managers and care package coordinators.
This cohort carries a 56% burnout rate. They’re managing a diverse workforce that’s often casual and agency-based. And when an AI project lands, they’re typically expected to contribute around 20% of their time to it, on top of an already demanding day job.
“They are sandwiched between demands from the executives and the reality of everyday work,” says Ali. “Your typical training isn’t going to be sufficient. We talk about the need for proper coaching.”
The risk is that this group checks the boxes. Yes, they attended training, yes, they can use the tool, but they don’t own it. And ownership matters.
Giving your middle managers a seat at the steering committee, with the authority to escalate issues, changes the dynamic. It signals that their participation isn’t a nice-to-have. It’s critical to success.
Have you protected their time to actually contribute? If not, that’s worth sorting before go-live, not after.
Frontline workers need to actually experience AI
Here’s the thing about frontline care workers: you can’t convince them with a slide deck.
“It’s not a cohort that you can basically convince,” says Ali. “They need to be able to see it. They need to be able to feel it, and they will make that judgement.”
Ali Kaabi speaking about frontline workers at the recent webinar, Implementing AI: Moving your team from ideas to impact
There’s also a genuine fear that needs to be acknowledged, the worry that AI is a monitoring tool, or that it’s there to replace jobs rather than support them. Saying “AI is here to help, not replace you” isn’t enough. Workers need to experience that truth in their daily work.
And that experience only comes through habit.
Habit is the missing piece
Thilan Perera, SVP, Customer Success at AlayaCare, has overseen many AI implementations. His observation is consistent: the technology usually isn’t the hard part.
“The hard part was actually building new routines,” says Thilan. “Until you start to embed this new AI tooling into daily routines and make it a habit, these AI tools — as great as they are — will unfortunately become one of those things that you set and forget.”
One customer from AlayaCare’s beta program put it simply: “The hardest part was the habit. Once it was adopted, it became indispensable.”
That shift from “something we tried” to “something we can’t work without” doesn’t happen through a one-off training session. It requires deliberate effort, sustained support and someone in your organisation who owns the adoption journey beyond go-live.
What you can do differently
If you’re planning an AI rollout — or trying to rescue one that hasn’t landed well — here’s where to focus:
- Track adoption by cohort, not overall numbers. Don’t measure uptake as a single figure. Track separately across your admin team, middle managers and frontline workers. Know where you’re winning and where you’re losing ground.
- Plan for backfill. If key managers need to contribute to the project, protect their time. Have a direct conversation with your executive sponsor about how you’ll cover their day job while they’re involved.
- Nominate an adoption owner. This isn’t a new hire. It’s an existing, respected person in your organisation who becomes the internal champion for AI adoption. They matter most at go-live and in the 30 to 90 days that follow.
- Keep the support going past day 30. The real adoption journey is only beginning at go-live. Make sure your frontline workers have access to help long enough for new habits to actually form.
The adoption gap between admin and frontline staff isn’t inevitable. But it doesn’t close on its own.
AlayaCare’s AI solutions are designed to support care organisations throughout every stage of implementation, from data readiness to long-term adoption. Talk to our team about what AI adoption could look like for your organisation.
Yes. In most aged care AI rollouts, administration and finance teams adopt quickly while frontline care workers lag behind. Strong overall adoption numbers can be misleading if they’re driven entirely by back-office staff. The key is to track uptake separately across each cohort — admin, middle managers, and frontline workers — so you know exactly where engagement is falling short.
Care managers and facility managers carry a 56% burnout rate before an AI project even begins. They’re typically asked to contribute around 20% of their time to a rollout, on top of an already demanding workload. Without protected time and genuine authority, they tend to check the boxes — attending training, learning the tool — without ever truly owning the adoption.
Frontline care workers can’t be convinced with a slide deck — they need to experience AI value firsthand. Providing hands-on exposure in real workflows, rather than one-off training sessions, is what builds genuine buy-in. It’s also important to directly address the fear that AI is a monitoring or replacement tool; workers need to experience it as a support for their day-to-day work, not a threat.
The technology is rarely the problem — building new routines is. Until AI is embedded into daily workflows and becomes habit, even effective tools get set and forgotten. As one customer from AlayaCare’s beta program put it: “The hardest part was the habit. Once it was adopted, it became indispensable.”
The critical window is the 30 to 90 days after go-live. This is when new routines either take hold or fall away, and when frontline workers most need accessible, ongoing support. A single training session at launch isn’t enough — sustained coaching beyond day 30 is what separates successful AI adoption from tools that get abandoned.
Yes — though it doesn’t need to be a new hire. An adoption owner is an existing, respected member of your team who becomes the internal champion for AI uptake. They’re most critical in the weeks following go-live, when habits are forming and frontline workers need a trusted point of contact rather than a helpdesk ticket.
Rohling recommends focusing on four things: track adoption by cohort (not as one overall figure), protect key managers’ time by planning for backfill before go-live, nominate a dedicated adoption owner, and keep support running well past day 30. The gap between admin and frontline AI adoption doesn’t close on its own.
