Michi — A reason to step outside

The description states that Michi is an iOS app built with Swift and SwiftUI. It uses GPT-5.6 for structured interpretation of user intent, MapKit for route generation, and Tokyo open data for personalization. The app allows users to input vague walking intentions (e.g., “I want greenery and fewer crowds”) which are converted into constraints by the AI. These constraints are then used to evaluate real MapKit routes based on public data.

• The product is described as a prototype, not a commercial offering.
• It uses real Tokyo public data for route personalization, but does not generate or invent destinations or roads.
• The AI (GPT-5.6) is used only for intent-to-constraint translation, and never receives user location history, usage metrics, or candidate routes.
• The app is designed to be privacy-preserving, with on-device storage of journals, feedback, and route history.

Not evidenced No evidence of revenue, customers, or commercial adoption. No information about how the product would scale beyond a single city or data source.

The description states that Michi is designed to help people who “struggle with the decision to go outside when there is no reason to do so.” It aims to give walking a small, curiosity-led purpose without gamification (e.g., streaks or points).

• The app positions itself as non-gamified, explainable, and curiosity-driven.
• It claims to use real data and AI for intent interpretation, not for route generation or decision-making.
• The product is described as a personalized walking assistant, not a safety tool, accessibility aid, or real-time congestion monitor.

Not evidenced No evidence of market positioning beyond the hackathon submission. No claims about user acquisition, retention, or monetization.

The author states that Michi is for people who “do not dislike walking; they struggle with the decision to go outside when there is no reason to do so.”

• The target customer is described as someone who lacks motivation to walk but could be nudged by a curiosity-driven prompt.
• It is not described as targeting specific demographics, accessibility needs, or safety concerns.

Not evidenced No evidence of user personas, segmentation, or market research. No indication of whether the team has validated this customer need beyond their own assumptions.

The description states that Michi is a prototype, not a commercial product.

• It does not mention any pricing model, subscriptions, or monetization.
• The app is described as not a safety tool or real-time congestion guarantee, implying no high-stakes use case or premium features.
• No evidence of revenue streams, partnerships, or B2B applications.

Not evidenced No business model, pricing structure, or commercial strategy beyond the prototype stage.

The project is built with Swift, SwiftUI, MapKit, and uses Python for backend logic. It includes:

• A local proxy to validate GPT-5.6 outputs.
• Automated tests: 208 checks (154 unit, 24 UI, 30 Python).
• Privacy hardening, including no API key bundling in the iOS app.
• The app is designed to work without an API key for core functionality.

• It uses Tokyo open data and MapKit for route generation.
• The AI is used only for structured intent interpretation, not for generating or scoring routes.
• The system supports deterministic routing and fallback behavior if the proxy is unavailable.

Not evidenced No evidence of scalability beyond a single city (Tokyo), or of how the system would handle more complex data sources or larger user bases.

The project is described as a hackathon prototype, submitted to the OpenAI 2026 hackathon.

• No evidence of revenue, customers, or usage metrics.
• No mention of product-market fit, user feedback, or iterative development beyond the prototype stage.
• The app is described as not a safety tool or real-time system, suggesting it’s not yet in production use.

Not evidenced No traction data, user adoption, or growth indicators. No evidence of product maturity beyond initial design and testing.

The description does not mention any direct competitors.

• It is described as a curiosity-driven walking assistant, not a fitness tracker, navigation app, or safety tool.
• It uses Tokyo public data and MapKit, which are widely available tools in the iOS ecosystem.
• No evidence of competitive differentiation beyond its use of AI for intent interpretation and public data personalization.

Not evidenced No competitive analysis, market positioning, or awareness of similar products.

• The app is described as a hackathon prototype, not a product in development.
• No evidence of commercial viability, scalability, or long-term strategy.
• The use of GPT-5.6 for intent interpretation is limited to structured output, but the AI is not trained or updated.
• The system relies on Tokyo-specific public data, which may limit its applicability.
• No evidence of user feedback loops, product iteration, or market validation.

Inference If this remains a prototype, it may not be ready for commercialization or investment.

1. What is the team’s plan to move beyond the hackathon prototype?
2. Are there any plans to expand beyond Tokyo or use other public data sources?
3. How would you validate user need and adoption beyond the prototype stage?
4. Is there a long-term vision for monetization, partnerships, or product development?
5. What are the technical challenges in scaling this system to more cities or users?