Inspiration
Most AI is built to give people answers. Rolling Thunder began with a different question: What if AI could help people understand what they are living through—without deciding what their experience means for them?
In my work, I have learned that real guidance begins by meeting people where they are. Before you can help someone see a situation differently, you need to understand what they are seeing from where they are standing.
Rolling Thunder is a digital application of the Stormwalker Method, a philosophy of guided independence. The goal is not to create dependence on an AI or position it as the authority. The goal is to strengthen self-understanding, perspective, resourcefulness, and self-trust.
What it does
Rolling Thunder is an AI reflection companion powered by the Stormwalker Method.
The user begins by choosing the kind of storm they are walking through. Instead of immediately offering advice, Rolling Thunder asks what that storm looks like in the user’s own words.
It then learns how the person naturally understands the world—the experiences, systems, crafts, environments, or forms of language that already make sense to them.
The Bridge Walk explores what keeps happening, what the person fears may happen if they stop responding in the same way, and what the pattern is costing them.
Before translating anything, Rolling Thunder reflects back what it believes it is hearing. The user can correct or confirm that reflection. Only after confirmation does the system translate the insight into a metaphor grounded in the user’s own language.
The experience ends with one realistic next move and a Lantern message that returns authority to the user.
Rolling Thunder does not diagnose, rescue, replace therapy or medical care, or claim authority over someone’s life. It is designed to help people recognize their own patterns and find their own footing.
How we built it
Codex helped turn the Stormwalker Method into a working web prototype.
GPT-5.6 powers the live conversation, while code-controlled stages guide the experience through a deliberate sequence:
- Choose the Storm
- Learn Their Language
- Bridge Walk
- What I’m Hearing
- User correction or confirmation
- Translation
- One realistic next move
- Lantern ending
The controlled stages matter because the order is part of the Method. They keep the AI from skipping ahead, making assumptions, or trying to solve a person before it understands them.
The prototype combines structured conversation design, carefully defined stage instructions, user verification, and live AI responses. The code controls when the experience may advance, while the model handles the natural conversation within each stage.
Challenges
The greatest challenge was not getting the AI to respond. It was teaching it when not to respond with an answer.
AI naturally tends to infer, advise, reassure, and move toward a solution. Rolling Thunder needed to slow that process down. It had to learn the person’s language first, reflect what it heard, allow correction, and wait for confirmation before translating the experience.
Another challenge was preserving the humanity of the Stormwalker Method inside a structured digital experience. The stages needed enough control to protect the philosophy without making the conversation feel mechanical.
Accomplishments
I am proud that the finished prototype demonstrates the complete reflection and translation process rather than only presenting the idea.
It can learn language from a person’s lived experience, use that language meaningfully, reflect its understanding for verification, accept correction, and produce a personalized translation without taking authority away from the user.
Most importantly, the prototype proves that AI can be designed to strengthen guided independence instead of encouraging dependence.
What we learned
We learned that personalization is not simply remembering facts about someone or changing the tone of an answer.
Meaningful personalization requires understanding how a person organizes experience and what language already carries meaning for them.
We also learned that user correction is not an interruption in the process—it is part of the intelligence of the process. Rolling Thunder becomes more accurate by giving the user a clear place to say, “That is not quite it,” before the system moves forward.
What’s next
The next step is to expand the prototype carefully while preserving its safeguards and philosophy.
Future development may include additional storm paths, stronger continuity across the reflection process, accessibility improvements, and more ways for users to save or revisit insights they choose to keep.
The long-term vision is not an AI that tells people how to live. It is a reflection companion that helps people understand what they are living through, recognize the medicine inside the storm, and trust themselves to take the next step.
Built With
- codex
- gpt-5.6
- netlify
- openai
- react
- typescript
- vite
- vitest
Log in or sign up for Devpost to join the conversation.