These days, almost anyone can build an AI chatbot in an afternoon.

Choose a platform. Upload a few documents. Write some instructions. Pick a color scheme. Add a friendly little icon. Connect it to a large language model and suddenly you have an “AI-powered” product.
There is nothing inherently wrong with that. Some of these tools can be useful. But a chatbot that can answer questions about hypnotherapy is not the same thing as a system that understands how hypnotherapists actually work.

And I think that distinction is about to become very important. When I started building Elah a few years ago, the instant AI bot ecosystem we see today didn’t exist in the same way. There was no convenient template for what I was trying to create because, frankly, I wasn’t trying to create a chatbot. I was trying to solve problems I had experienced firsthand as a working hypnotherapist.

I knew what it felt like to squeeze consultation calls between sessions. I knew what it was like to collect information through forms and still finish reading them with more questions than answers. I knew how much valuable information could emerge during a pre-talk, only to become scattered across handwritten notes, recordings, follow-up messages, and memory.

I had also spent years trying to solve completely different problems: How to effectively track progress without disrupting the life of my clients. How to create truly personalized hypnosis audios at scale and so on. Elah was born out of real pain and overwhelm and once I started the process of researching the market and validating the project, I discovered that not only were there no hypnotherapy focused platforms that solved the case management issues, there were no AI powered hypnotherapy platforms that could help solve even the most basic workflow needs of our community.

A List of Questions Is Not a Conversation

As hypnotherapists, we understand that the presenting problem is often only the beginning of the conversation. That is why I couldn’t simply build a form, attach AI to it, and call it intelligent.
Traditional intake forms ask predetermined questions. More sophisticated digital forms use conditional logic. If a person answers A, show question B. If they answer C, skip to question D.
Many AI chatbots are more flexible than that, but flexibility alone does not create meaningful intelligence.
A system can generate endless follow-up questions and still ask the wrong questions.
It can sound warm without knowing what matters. It can summarize a conversation beautifully while completely missing the pattern a practitioner would care about. It can produce a polished response without understanding where that information belongs in the larger client journey.

This is one of the biggest misconceptions I see around AI. People often confuse fluent language with deep understanding of a professional workflow. Large language models are extraordinarily powerful, but the model itself is not the profession.
It has not spent years sitting across from clients. It has not experienced the moment when someone books for weight loss and, halfway through the conversation, reveals that food is serving a completely different emotional function. It has not had to decide what information matters before a first session, what should be revisited later, what needs clarification, what belongs in a case record, or what should remain firmly within the practitioner’s professional judgment. You cannot solve that simply by telling an AI to “act like an expert hypnotherapist.”