Mythicself

This is not a productivity app. It helps you notice your myth.

Comment

This is not a productivity app. It won’t help you achieve your dreams. It helps you notice your myth.

Inspiration

In a world obsessed with optimization, efficiency, and "getting things done," our inner lives are often treated as mere data points to be managed. We noticed a significant gap in the mental wellness landscape: dream interpretation apps either act as unscientific "fortune tellers" (predicting the future) or dry "analyzers" (labeling your subconscious with static tags). We were inspired by the depth psychology of Carl Jung and the mythological frameworks of Joseph Campbell. They taught us that our dreams are not random noise, but chapters in a continuous "Hero's Journey." We wanted to build a digital sanctuary that uses AI not to give answers, but to hold up a mirror—helping users recognize the archetypal patterns weaving through their lives.

What it does

MythicSelf is a reflective AI companion that transforms fragmented dream journals into a cohesive personal mythology. It is "Netflix for your subconscious"—instead of consuming someone else's story, you wake up every morning to a new episode where you are the protagonist.

Core Features:

  1. Archetypal Mirroring (Not Analysis): Instead of diagnosing you, it identifies which universal archetypes (The Hero, The Shadow, The Wise Guide) are currently active in your subconscious.
  2. The Narrative Arc: It connects the dots between dreams over time. It remembers what you dreamt last week and explains how today's dream is a continuation of that story (e.g., moving from "The Call to Adventure" to "The Road of Trials").
  3. Reflective Dialogue: It acts as a Socratic mirror. It doesn't just tell you what a symbol means; it asks profound questions like, "You encountered a blocked door. What is currently inaccessible to you in your waking life?" empowering the user to find their own meaning.

How we built it

We built MythicSelf using Next.js 14 for the frontend and Google Gemini 3 Flash Preview API as the core intelligence engine. The project relies heavily on three specific capabilities of the Gemini 3 family:

  1. Long Context Window (1M+ Tokens): This is the heart of our "Narrative Arc" feature. Unlike standard LLMs that treat each prompt in isolation, we feed weeks of dream entries into Gemini's context window. This allows the model to "remember" recurring motifs (e.g., a specific blue bird appearing across different dreams) and construct a longitudinal psychological profile.
  2. Advanced Reasoning: We utilize Gemini’s reasoning capabilities to map abstract dream imagery to specific stages of Joseph Campbell’s Hero’s Journey. The model must justify why a specific dream represents a "Refusal of the Call," providing evidence from the text.
  3. System Instruction Tuning: We spent significant time engineering the system prompt to prevent the AI from being "prescriptive." We tuned the model to adopt the persona of a "Jungian Mirror"—using open-ended, non-judgmental language.

Challenges we ran into

  • The "Fortune Teller" Trap: Early versions of the model tried to predict the user's future or give superstitious interpretations. We had to rigorously refine the prompts to focus strictly on psychological reflection and "here-and-now" awareness.
  • Balancing "Mystic" vs. "Tech": We wanted the UI to feel like an old mythic tome or a tarot spread, not a SaaS dashboard. Implementing this "ethereal" feel using CSS and animations while keeping the app responsive was a design challenge.

Accomplishments that we're proud of

  • The Counter-Positioning: In a hackathon focused on speed and function, we are proud of building a tool that encourages slowing down. Positioning it as "Not a productivity app" sparked curiosity and resonated with our beta testers.
  • Coherent Narrative Generation: Seeing Gemini 3 successfully link a dream from Day 1 with a dream from Day 7 to tell a unified story was a huge "Aha!" moment. It proved that AI can handle long-term narrative continuity effectively.

What we learned

  • AI as a Mirror, not a Master: We learned that AI is most powerful when it acts as a co-pilot for self-discovery rather than an authority figure providing "answers."
  • The Power of Context: The difference between analyzing one dream and analyzing ten dreams together is massive. Gemini 3's context window opened up a new dimension of "pattern recognition" that wasn't possible with previous models.

What's next for mythicself

Multimodal Dream Reconstruction: We plan to integrate Gemini's video generation capabilities (Veo) to visualize dream scenes. Imagine describing a "Forest of Clocks" and watching it come to life.

  • Waking Life Integration: Expanding beyond sleep to include waking journals, allowing users to see how their dream archetypes manifest in their daily reality.
  • The Book of Myth: A feature to export your monthly journey as a beautiful PDF book or a printable Tarot card set.

Built With

Share this project:

Updates