MINDPATH

MindPath Landing Page
Onboarding & Personalization
AI Ethics & Transparency
Informed Decision-Making
Chat room with 0% (no context)
Chat room with 70% (full context)
All document of project can be downloaded in pdf or docs
PDF document
Word document

Inspiration The AI Project Thinking Partner (Second Brain) was born from a critical observation: student entrepreneurs are often trapped by "analysis paralysis," drowning in infinite research while losing sight of the core problem. We witnessed brilliant projects abandoned simply due to a lack of clear, methodical guidance. We built this system not as a passive chatbot, but as a proactive mentor designed to strip away uncertainty and replace it with a structured path to execution.

What we learned We realized that general-purpose LLMs struggle with "Prompt Drift"—they become complacent "yes-men" that lose track of the user's initial constraints. We learned that for an AI to be a true partner, its "memory" must be strictly decoupled from the chat history and stored in a persistent, logical state.

How we built the project Our architecture solves the context-loss problem through a Dual-Channel system: Visual Channel: A conversational interface for natural user interaction. Structural Channel: A State JSON Memory that functions as the project's single source of truth. By forcing the reasoning engine to process only this immutable JSON state, we render hallucinations technically impossible. We measure project readiness through a dynamic Global Comprehension Score ($S_{global}$), ensuring the project is validated across four pillars: Problem, Technique, Budget, and Risk.

Challenges we faced The biggest challenge was avoiding the "passive AI" trap. We solved this by engineering an intelligent, adaptive questioning system that performs logical entity extraction. Instead of repeating questions, the AI uses a "Micro-Insights" loop to provide immediate value before asking the next targeted question. Finally, we bridge the gap to reality by generating three immediate, actionable deliverables: a Dynamic Project Canvas, a Risk & Blind Spot Matrix, and a 30-Day MVP Roadmap.