Inspiration
At 22, I was already running a business I had started at 18, manufacturing detergents and ethnic cosmetics sold to wholesale cash-and-carry businesses across Southern Africa and generating several million dollars a year.
Then, travelling from Durban to Johannesburg, I was involved in a serious car crash that left me with what doctors described as an 80% brain disability.
That experience changed the direction of my life.
As I rebuilt, I realised that although circumstances differ, many human challenges are universal: relationships, fear, identity, money, purpose, confidence, resilience and change. I went on to earn two degrees, including a postgraduate finance qualification, become a certified banker and certified financial planner, and work in investment banking, commercial banking and wealth management.
After 17 years in corporate life, I returned to human potential, leadership and transformation. That became Elysium, a conventional leadership and human-potential business built around my book Elysium Odyssey, executive coaching, four books, multimedia programmes, leadership retreats and keynote speaking.
I later built Elysium One, a web and Android platform consolidating appropriate Elysium resources into one member library. But there was a problem: the library was broad, yet people still had to work out which resource they needed.
Then I began experiencing extraordinarily vivid dreams — airports, aircraft, water, shareholders' meetings and world leaders. I became fascinated by one question:
What if personal growth could begin with what a person's own lived experience was already revealing?
That became DreamCatcher.
What it does
Elysium DreamCatcher is a new AI product built on the existing Elysium business and Elysium One ecosystem.
DreamCatcher uses Gemini to turn a dream into a personalised journey:
Dream → Pattern → Insight → Personalised Resource → Action → Longitudinal Learning → Transformation.
It analyses symbols, emotions, movement, relationships, locations and patterns while incorporating personal context and previous dreams. It can then map emerging needs to relevant Elysium resources covering Personal Mastery, Mind Mastery, emotional resilience, relationships and communication, spirituality and grounding, purpose, wealth and money psychology, entrepreneurship and leadership.
As dreams accumulate, DreamCatcher can identify longitudinal patterns and show how those patterns appear to evolve.
This is why DreamCatcher matters in Education & Human Potential: it changes personal development from a fragmented, generic “find the content you need” model into an AI-personalised journey that begins with lived experience and adapts as a person evolves.
This connects directly to our Massive Transformative Purpose: Help, heal and transform one million lives by 5 p.m. on December 31, 2035, transforming people from heedlessness to wakefulness, ordinary to extraordinary, and working class to world-class, while empowering them to live more meaningful, measurable, and magnificently sustainable lives.
How we built it
We first created a lightweight GPT-wrapper version with a deliberately limited taxonomy and put it in front of paying customers to validate demand.
That validation gave us the confidence to build the deeper product.
Our major intellectual asset is the Elysium DreamCatcher Taxonomy — a proprietary interpretation and transformation framework developed from approximately 18 months of intensive dream observation, research into sages and thinkers, personal-development methodology and iterative testing.
Gemini is the intelligence layer; the Elysium taxonomy, coaching methodology, ecosystem and longitudinal framework provide the distinctive intellectual and commercial layer.
I am not a software developer. At 50, and while recovering from an 80% brain disability, I learnt about AI.
AI became part of the development model itself. I used AI to translate years of domain knowledge into architecture, coding, debugging, testing, deployment and iteration.
We rebuilt the production system using Google Cloud and Gemini through Google AI Studio, connected through GitHub, Git and Visual Studio Code, deployed through Render and embedded into Shopify. The production version went live during the final days of the Gemini campaign.
The architecture keeps API credentials server-side, with authentication and security controls including 2FA and inactivity/session controls.
Challenges we ran into
The hardest challenge was not getting AI to generate text. It was getting it to produce meaningful, personalised reasoning rather than generic dream-dictionary output.
We had to solve for personal context, structured taxonomy, resource mapping, longitudinal patterns and an experience that feels like an AI coach rather than a chatbot — while building production technology without a conventional development team.
We also learned that the international infrastructure of a digital business is not borderless. We incorporated a UK company to access international markets and payment infrastructure, encountered banking and regulatory obstacles, and ultimately adapted our structure and payment infrastructure back to South Africa.
The biggest lesson was simple: AI capability alone is not the product. Defensible value comes from combining AI with distinctive IP, domain knowledge, methodology, customer feedback and a real business model.
Accomplishments we're proud of
We moved DreamCatcher from concept to a live, revenue-generating product during the hackathon.
We validated demand before overbuilding, then deployed a Gemini-powered production system connected to Shopify. Customers can now access personalised dream interpretation, follow-up dialogue, journaling, longitudinal insights, pattern recognition and personalised connections into Elysium One.
We are particularly proud of the AI-coaching model: Elysium One contains a broad treasury of human-potential resources; DreamCatcher becomes the intelligent layer that helps determine what may be relevant to a person at a particular moment.
We also won an AI Publishing Award at the London Enterprise Awards.
What we learned
We learned to ask not “What can AI do?” but “What should AI accomplish for the customer and the business?”
We learned to validate before overbuilding, and that proprietary methodology matters. The model provides extraordinary capability; our taxonomy gives that capability a distinctive purpose.
Most importantly, we learned that AI changes who can build. I entered as a founder, strategist and domain expert rather than a software engineer, yet AI helped me translate years of knowledge and IP into a functioning production product.
That is the opportunity of AI-native entrepreneurship: a small organisation can achieve disproportionate impact through technology.
Category Impact
DreamCatcher is designed to make personalised human-potential education accessible at scale. Instead of asking millions of people to navigate fragmented content, AI can help each person discover what is relevant to them, from their own lived experience.
Our impact hypothesis is that greater self-awareness, pattern recognition and personalised action can become catalysts for longer-term human development. We recognise that sign-ups and DAUs are not sufficient measures of transformation; establishing robust measurable proxies for outcomes remains an important next step.
What's next
We will deepen longitudinal intelligence, improve voice interaction, and better customise the ritual exercise. (Judges: please note that these two features are currently in beta and may not always work as intended. We will address and improve them after the winners are announced, in accordance with the rules governing changes to code during the judging period.)
We will also expand personalised resource mapping and establish stronger evidence of outcomes over time.
We are investigating additional IP protection for DreamCatcher, in addition to the three trademarks and other intellectual property assets we already hold covering our broader Elysium assets.
Dreams are the doorway. Transformation is the destination.
That's the story I'm into the judging room in Los Angeles on September 25, 2026. 🔥
Built With
- chatgpt
- github
- google-ai-studio
- google-cloud
- google-gemini-api
- invideo
- node.js
- react
- render
- sidekick
- typescript
- vite
Log in or sign up for Devpost to join the conversation.