Meet Peggy: the AI Concierge Working for Staff and Residents in a Later-Living Community
If you manage, operate, or oversee a later-living development, you already know the feeling of wearing twelve hats at once. There's a leak in Flat 22, a contractor who's running late, and three residents waiting to ask you questions about the recycling, a parcel and the time of the yoga class.
In theory, technology is supposed to help. In practice, it often adds just one more thing to manage. More portals, more passwords, more PDFs no one reads.
This is about a different kind of tech. One that's already working usefully in a community like yours. It doesn't need a tutorial. It doesn't care what time it is. It answers questions in plain English, out loud.
It’s a voice-enabled AI chatbot. We call her Peggy. Of course, you can choose your own name.
A short personal diversion (bear with me)
I've been early to a few tech changes in my career. I published some of the UK's first computer games before IBM had PCs on desktops. Later, I helped businesses make sense of email marketing at a time when checking your inbox more than once a week was still a novelty.
I'm not a 'techie'. I've simply been intrigued by the intersection between people and new technology.
So, when I moved into a newly built over-60s development right around the time AI went from "interesting" to "usable", I saw another opportunity ideally suited to my generation.
Why the problems keep multiplying
From the inside, I could see how these developments operate. Staff stretched by property management of complex systems never knowingly under-engineered. Residents asking the same perfectly reasonable questions, again and again.
The real problem? Not the people. The systems.
Information is often there, buried in a handbook, on a noticeboard, or in someone's head. But the effort of finding it is frustrating and disempowering for residents. That's not what later living is supposed to feel like.
That's when it clicked. This isn't a staffing problem. And it's certainly not a resident problem. This is a systems problem that AI, specifically voice-enabled chatbots, might actually solve.
Your Team to the Power of AI
"Peggy" doesn't write poems. Won't tell you who won "Strictly" in 2006. And won't replace a human, either. What it does is answer residents’ questions reliably, patiently, and instantly.
"How do I test my smoke alarm?" "Is the coffee morning still on?" "Who do I call if I haven’t got hot water?"
No waiting around. No holding music. No flipping through handbooks.
It's AI for the real world, tuned to your development, your systems, your language. Your team to the power of AI. AI to support, not supplant the humans.
And yes, it works. In our pilot, staff were soon suggesting residents "just ask Peggy". And residents? They described her as "unflappable and invaluable". One even said, "Looks like AI will take care of us in our old age". (I'll take that as a review.)
The important bit: what makes it work (and why it’s harder than it looks)
None of this happened by magic. If you've tried ChatGPT or other large language models, you'll know they can help you write an email or plan a trip. You also probably know they might invent a few facts while they're at it. Like a bright, slightly unreliable intern after a night out.
Building something residents trust takes more than pointing a chatbot at a set of documents and hoping for the best.
We layered large language models, speech interfaces, prompt structures, and data curation. Then we tested. Extensively. Real questions, real residents. Structured evaluations. We watched what made people hesitate. What made them smile. What made them trust the answer.
And we went further.
What happens if someone says, "I'm not feeling well today"? Or uses a phrase no one's heard since 1953? Or interrupts themselves halfway through?
These are not edge cases. They're real life. Which means Peggy has to respond like something more than a search engine with a voice.
In many ways, training our AI chatbot reminded me of "proofing" in dog training. (I’ve written before about the parallels in training a companion – whether AI or dog.) Teaching a dog a cue is one thing. But will they still sit when it’s raining, the handler’s distracted, or a squirrel’s racing across the street? That’s the real test. You’re not just checking knowledge. You’re testing consistency in the face of distractions and ambiguity.
Peggy won't fetch a ball, but she's good at making friends and building trust.
Let's talk if this sounds familiar
If you're running or overseeing a later-living development, I'm guessing some of this rings true:
Good staff, stretched thin
Repetitive resident queries
Information scattered across too many channels
And on a more practical level, you're dealing with more and more moving parts. Emergency call systems to test. Sprinklers and backup generators to maintain. Electronic gates, lifts and smart-home devices that need regular servicing – and raise frequent questions from residents.
What you might need is something simple, solid and human-friendly. Something that takes pressure off staff, supports residents, and works quietly alongside your existing set-up. Something that sits on top of your workflows with no heavy training required, no extra headaches.
That's what we're building. With people like you, for people like you. Giving time back to staff for more fulfilling conversations.
Remember your residents are a generation that's lived through more technological change than any other in history. These are people who've adapted from rotary phones to smartphones, from handwritten letters to email, from vinyl to streaming. They don't need patronising; they need technology that respects their intelligence and doesn't waste their time. They may not all know what a QR code is, but they definitely don't want to feel silly asking.
If you're curious – or even just sceptically interested – I'd love to talk.
We can even let Peggy join the conversation.
