A few weeks ago I posted a video created by Pantio, a company I advise. Pantio creates AI-powered digital personas of deceased loved ones. Families can interact with them by voice or text, hear their stories, ask the questions they never got to ask while there was still time.
One of my favorite people in this profession (and in the world) commented, “Sir this is all ai thooooooo”
Another wrote “Agree. Unhealthy 🤮”
I wrote back: Yes, of course it is. People replay voicemails from people they’ve lost. They reread old letters. They talk to headstones. Nobody calls any of that unhealthy.
I’ve listened to the voicemails, read and reread the text messages. Everyone in my family has died, most passed fairly recently. Those recordings are connection. They’re proof that someone existed and that they loved me. The only difference between a voicemail and what Pantio does is quality and interactivity. The grief is identical. The need is exactly the same.
The question isn’t “is this AI?” The question is whether it serves the family.
Our profession has never been good at asking the second question first.
The pattern is older than AI
The hand-on-the-shoulder photo. The bowed head. The candlelit tableau staged for the feed. The condolence email that sounds correct and lands with a thud. The obituary that uses “beloved” and “cherished” because it’s faster than asking the family the harder questions.
None of that started with AI.
These patterns lived in funeral marketing for two decades on social media and longer than that in print. They got pushed into the work by competitive pressure, by a consumer culture that rewards image over substance, and by a public that started judging funeral homes on aesthetics long before any of us could spell GPT. The pressure to look caring instead of be caring was already eating at the profession.
AI scaled that pressure and made it cheap. But the pattern was already there.
This matters because it changes the assignment. If AI is the cause, you fight AI. If AI is the mirror, you fix what it’s reflecting. Banning the tools doesn’t address what was already shifting beneath the surface. It just slows things down while the same pressures keep shaping the work.
The objection worth taking seriously
The same person who commented on that post is also a celebrant, and a good one. Which means she knows better than most what it looks like when the human process gets stripped out of service work.
A celebrant-led service isn’t a eulogy in the traditional sense. The celebrant doesn’t know the deceased. She knows them through the family, through hours of interviews that surface who this person actually was and what the people who loved them need to say. Then she goes home and builds something from what she heard, making careful decisions about what to hold and what to leave out. That process is the work. If AI generates the service instead, the family receives something that looks like the product of that labor without any of it having happened. No one sat with them. No one listened. The output might read fine. Everything that was supposed to precede it never existed.
That line is worth holding. And she’s right to hold it.
But here’s where it gets interesting. The same celebrant who will never let AI write a service might find that AI transcribing her family interview, so she can be fully present in the room instead of taking notes, is one of the most useful tools she’s ever touched. One use puts a machine where a human is supposed to be. The other clears the runway so the human can do what only a human can do.
That’s the distinction the “but it’s AI though” reaction tends to skip.
AI doing the work versus AI doing the chores
This is the only distinction that matters, and it keeps getting lost. And for the record, AI also fixed my spelling in this piece and killed the run-on sentences I was apparently so well known for. Anyone who’s gotten a midnight text from me knows that last part was no small lift.
AI doing the work means the service gets written by a machine. The condolence email gets generated. The candid moment gets fabricated. Those uses put synthetic output in the place where presence is supposed to live. The concern is legitimate.
AI doing the chores means something else. It means pulling structured data from an arrangement conference recording so the director isn’t writing while a widow is talking. It means handling back-office documentation that burns three hours at the end of already-long days. It means drafting routine compliance correspondence that has nothing to do with the family, so the person who should be on the phone with the daughter who keeps calling back can actually be on that call.
That’s not synthetic compassion. That’s what gives a stretched-thin, burned-out profession back the hours it needs to practice the real kind.
You can’t build a humane profession on exhausted staff and patch the gap with software. You also can’t build a humane profession on exhausted staff and refuse to use software at all. The hours have to come from somewhere.
The question worth asking
This profession has been here before. Cremation was going to cheapen death. Celebrants were going to hollow out meaning. Online pricing transparency was going to commoditize everything. In every case, the first response was emotional and categorical. The useful things eventually got integrated, the harmful things got filtered out, and the profession moved forward.
AI is that conversation again. The people who stop at “but it’s AI though” never get to the useful part. They’re treating a category as a verdict.
The right question is not whether technology is involved. It’s whether the tool puts a machine where a human is supposed to be, or clears the runway so a human can actually land.
One of those corrupts the work. The other one protects the hours that make the work possible.
That distinction is worth fighting about. Everything else is noise.