AI in Practice: three operators on what's working, what took real effort, and where to start
The morning's first breakout brought together three people actively running AI inside their organizations: Stephen Eatman, Vice President at Sunshine Retirement Living; Kerri Pyne, Regional Director of Sales at Sunshine; and Matthew Barger, Vice President of Sales and Hospitality at Maxwell Group. Sara Vozeolas, WelcomeHome's Director of Operations, moderated. The framing was deliberately practical. Everyone hears about AI constantly; this panel was about what implementation actually looks like, from people doing it.
The problems they chose
Both organizations started with a real bottleneck rather than a tool.
For Sunshine, it was two things. First, the annual refrain of "we don't have enough leads" had gotten stale, so the team flipped the question: what about the thousands of leads already in the database that aren't doing anything? Reactivating dormant leads became the year's central AI project. As Stephen put it to his sales teams, you don't need new leads. They're sitting there waiting for you to call them. Second, phones. With concierge staff pulled away and sales out on tours, roughly a third of inbound calls were dropping. Integrating an AI answering system took far more work than plugging something in, including rebuilding parts of the phone system, but the payoff was concrete: the drop rate fell from around 35% to 15%. In an industry where being first to respond matters, that's a meaningful number. Sunshine also rebuilt its website about two years ago specifically to perform well in AI-driven search, a year-long project that predated most of the current conversation.
Maxwell Group's focus has been moving from exploration to implementation, primarily with Microsoft Copilot. Matthew's most vivid example was hospitality prep: building a "life story" of each prospect from discovery notes and tour conversations, then using it to plan the visit. Not just "Sally is coming from Atlanta," but she just had a hip replacement and has been in the car for four and a half hours, so tell her to pull up rather than park, ask therapy about a seat cushion, and have an ice pack ready. His point was that AI can give team members who aren't naturally hospitality-minded a way to deliver that level of intentional care on every tour.
The tools they'd call non-negotiable
Asked what they'd recommend to peers, the panel got specific. Matthew pointed to WelcomeHome's call intelligence, particularly talk-to-listen ratio matched to a prospect's stage of readiness, and AI Snapshot, along with Microsoft Copilot, which he admitted he once dismissed before its agent capabilities won him over. Stephen credited the WelcomeHome partnership itself, especially the back-and-forth of pitching ideas and getting reporting that shows how AI channels are performing, plus Further, which powers Sunshine's chatbots and AI phone answering. Kerri's list ran toward people development: AI-assisted interviewing for concierge hires, an AI role-play partner for training new and seasoned salespeople, and feeding WelcomeHome reports into an AI assistant to pinpoint exactly what each team needs coaching on.
Getting teams to actually use it
Adoption was the most candid stretch of the conversation. Kerri described herself as an early foot-dragger, worried that AI would cost the industry its human touch, and said her teams shared the fear. What changed the conversation was the pitch. Not "here's a new tool I want you to use," but "here's a tool that will save you three hours." Once teams saw time saved and conversion improve, buy-in followed.
Two other adoption lessons stood out. Sunshine runs most of its AI behind the scenes, so communities mostly experience the results, and when a salesperson objects that the AI scheduled a tour they wanted to schedule themselves, the answer is simple: beat the agent to the lead. The reactivation agent only touches prospects with no activity, so any salesperson who follows up first wins. It's turned into healthy competition, and either way the prospect gets the call. Matthew added a discipline that applies to every AI-generated touchpoint: remember the receiver. The average buyer is in her early 80s, and everything from the message to the font size should be built for her, not for the tool.
The panelists were honest about the friction, too. Seasoned sales advisors are the most hesitant to change what's worked for decades, so start them small and low-risk. And the vendor landscape itself is exhausting; Stephen estimated that of the 40 vendor emails in his inbox, 15 are selling an AI tool, and there's always a newer version of whatever you just rolled out.
Advice for operators starting now
Three answers, all compatible. Start small: find one bottleneck, pilot a tool at a handful of communities, measure, and let the pilot participants sell it to their peers, because buy-in travels better peer to peer than top down. Plan for the new bandwidth before you create it: when a two-hour task takes 12 minutes, that time turns idle unless leadership articulates what it's for, and pointing it at more resident and prospect interaction is also a retention play, since people stay in jobs where they spend time on what they want to do rather than what they have to do. And start with your existing partners: tell them your problem before shopping the market, because several of Sunshine's longest-standing vendors were already building what they needed.
The guardrails, from audience Q&A
The audience pushed on governance, and the answers were unusually concrete. At Maxwell Group, every AI tool except Copilot is blocked on corporate computers, precisely so financials and sensitive documents never end up in public tools; Matthew's advice to anyone without a policy was to start writing one. Sunshine uses tiered licenses so executive-level material stays at the executive level. On brand control, Stephen described walking into a community on a site visit and spotting five AI-generated flyers before he reached the elevator; both Sunshine leaders route creative through marketing and use WelcomeHome's events tools to keep eyes on what goes out. And on chatbots, the feedback loop matters more than the pick: test it, try to break it, and send every community complaint straight to the vendor. One audience member offered a useful counterpoint from the CCRC world, where entry fees and complex pricing have held chat completion to 13%, a reminder that the same tool performs very differently depending on what it's being asked to explain.
The closing note was unanimous, and it echoed through every answer: this is a relationship business. AI does the heavy lifting and the preparation so the humans can have more of the interactions that actually move someone in.