How does Aphobic work?
You might assume Aphobic is just another app built on top of a ChatGPT prompt.
We could have asked ChatGPT to read Chris’s book, Face Your Fears, and respond as if it were him. But that’s not how real therapy works, and it’s not how Aphobic works either.
While we love ChatGPT and other Large Language Models (LLMs), there’s one thing they aren’t good at: following a structured process.
If you asked ChatGPT to act like Chris, it might sound convincing for a few messages, but it would quickly get off track, jump ahead, or miss important details.
To build something that truly reflects Chris’s therapeutic method, we had to take a different approach.
We manually modeled Chris’s 7-step process for relieving fears and phobias, working closely with him to understand:
what he says and when;
what he avoids;
how he makes decisions during a session;
which techniques he uses and how he chooses them;
and, how to recognize and respond to a client’s emotional state.
Aphobic uses a structured AI approach, powered by real therapeutic insight, so that it can guide you through your own journey with fear and phobias.
The 7 Steps of Integrated Change System Therapy
Know Your Fear
Relax the Conscious Mind
Uncover Secondary Gain
Discover & Scramble Your Strategy
Discover & Scramble Past Memories
Recondition Your Emotions
Face Your Fear
This is not just another AI chat bot
Breakthrough AI Approach
Aphobic is built on a breakthrough AI approach called subgoal-directed LLMs with committee.
The ultimate goal of the AI approach is to help relieve you of your fear or phobia using the ICS therapeutic process.
To follow such a process, we model it using a subgoal map. At each point in the conversation between the user and the AI, the AI chooses what the next subgoal is.
Essentially the subgoal approach is like a therapist asking themselves:
“Should I keep the client on this topic, or ask them how they’re feeling, or ask them how this relates to something they said earlier?”
To drive the choice of the next subgoal, we use a host of awesome tech like LLMs to handle intent recognition, parsing, and understanding user state.
Building Safe AI
To support emotional awareness, Aphobic includes a committee layer of AI agents designed to monitor patterns in the user-AI interaction. These agents track signals such as elevated distress, repeated avoidance, or signs of engagement difficulty.
While Aphobic is not a healthcare company, we take your privacy and security incredibly seriously. We follow data security best practices aligned with HIPAA standards.