You Can’t Prompt Your Way Into Lived Experience

A large part of building Elah came from my own experiences, both good and bad. Forms that were too long. Forms that became too short. People who gave incomplete answers. People who paid for personalized audios and still never completed the intake. Consultation calls that consumed hours.
Leads who turned out to be outside my scope. Information that was collected once and then became disconnected from everything that happened afterward.

These experiences were frustrating at the time, but I wasn’t alone. A lot of my colleagues were going through the same and I had to choose a handful of the most pressing problems. They became part of the intelligence behind the system.

Every failure taught me something about the actual problem.
The problem wasn’t that practitioners needed another form.
The problem wasn’t that we needed another generic chatbot.
The problem wasn’t even that we needed AI to write more scripts.
The deeper problem was fragmentation.

Client information enters a practice from multiple directions. Intake. Pre-talk. Session observations. Follow-up. Progress updates. Practitioner reflections. Case notes. Client language. Goals. Outcomes. Over time, the story evolves.

But most software treats those moments as separate events. A hypnotherapist doesn’t.

We carry the evolving story of the client in our minds. We remember what they said three sessions ago. We notice when their language changes. We compare what they say now with what they believed before. We recognize when a metaphor returns, when a goal shifts, when a pattern repeats, or when something that once triggered a strong response suddenly no longer does. That continuity is part of the work.

So if AI is going to support hypnotherapists meaningfully, it cannot simply sit in a chat window waiting for a prompt. 

Real Intelligence Requires Context
Imagine a practitioner writes this after a session:

“Client seemed more confident today.”
On its own, that sentence tells us very little.

But what if the client’s intake showed a long-standing fear of disappointing others? What if the first session notes documented difficulty making decisions without reassurance? What if a later progress update showed that the client finally set a boundary at work? What if the practitioner’s own reflection noted a visible shift in posture when the client described the experience?

Now we are no longer looking at a sentence. We are looking at movement. That is where AI becomes interesting to me. Not because it can pretend to be the therapist. Not because it can generate an impressive paragraph. Not because it can replace professional judgment. Because it can help connect information that already exists.

It can help organize fragmented observations, surface recurring themes, compare changes over time, and bring relevant context forward when the practitioner needs it.

But that kind of support requires more than a clever prompt. It requires thinking deeply about how information moves through the entire workflow architecture. Asking questions that are far from chatbot questions. They are system-design questions.

The Hard Part Isn’t Making AI Talk

Making AI talk or text back is easy now. Making it useful inside a real professional workflow is much harder. Building guardrails is incredibly complex given that we’re not only dealing with words, but also things like emotional and resistance detection. Then there is the matter of deciding what the AI should pay attention to, what it should ignore, when it should investigate further, how different pieces of information relate to one another, what should persist over time, and how to support the practitioner without attempting to become the practitioner. That last distinction matters enormously to me.

I do not believe the future of hypnotherapy is an AI pretending to be human. I don’t want AI replacing the pre-talk. I don’t want it manufacturing fake rapport. I don’t want practitioners becoming passive while a machine tells them what to do.

I want the opposite. I want the practitioner to walk into the pre-talk better prepared and more present.
I want them to spend less time trying to remember what they read on a form and more time observing the human being in front of them. I want post-session reflections to become useful later instead of disappearing into notebooks. I want progress tracking to reveal what is actually working. I want information gathered early in the client journey to remain meaningful throughout that journey. I want AI to strengthen human attention, not compete with it.