Inspiration
Big decisions don’t feel like checklists. They feel like standing at a fork in the road, unsure which path leads to growth and which leads to regret. While experimenting with AI tools, I noticed that most decision assistants return long blocks of text or generic pros and cons, which don’t reflect how humans naturally think about choices. The inspiration behind this project was to rethink decision-making as a visual and narrative process, not just an analytical one. I wanted to build an application where AI doesn’t simply answer questions, but maps choices as paths, helping users understand consequences before committing.
What it does
This project is a visual, path-based decision analysis assistant powered by the Gemini 3 API. When a user submits a dilemma, the system: Frames the problem as a crossroads Generates exactly three realistic paths forward Breaks each path into pros, cons, trade-offs, and long-term outcomes Asks clarifying questions before making any recommendation Recommends one path based on the user’s priorities Assigns a confidence score to reflect uncertainty and assumptions The analysis is designed to be rendered as a comic-style decision map, making complex choices easier to grasp and emotionally engaging.
How we built it
The core of the application is built around the Gemini 3 API, which acts as the decision reasoning engine. Designed a path-based decision framework inspired by visual storytelling Used Gemini 3’s reasoning and long-context understanding to: Interpret ambiguous dilemmas Compare multiple future paths Adapt recommendations across turns Enforced a strict structured output format so results could be rendered visually Built a lightweight interface that converts Gemini’s structured responses into comic-style panels Gemini 3 is central to every step — from framing the dilemma to evaluating trade-offs and adapting when new constraints are introduced.
Challenges we ran into
One major challenge was ensuring that Gemini consistently produced structured, UI-ready outputs instead of free-form text. This required careful prompt design and iterative testing. Another challenge was balancing creative storytelling with analytical depth. The responses needed to feel visual and human, without sacrificing clarity, reasoning quality, or decision rigor.
Accomplishments that we're proud of
Built a non-generic decision assistant instead of a chatbot Successfully used Gemini 3 to perform multi-path reasoning Designed an AI system that supports visual decision-making Created a format that makes complex trade-offs easy to understand Delivered a working, testable product within contest constraints
What we learned
Gemini 3 excels at reasoning when given clear structure and constraints Visual metaphors significantly improve how users understand AI outputs Strong prompt design can turn AI from a text generator into a decision engine User trust improves when AI explains why a path is recommended
What's next for Untitled
Future improvements include: Interactive “what-if” path simulations Support for collaborative decision-making Expanding into domains like career planning, startups, and product strategy Further refining Gemini prompts for deeper long-term outcome analysis This project is a step toward using AI not just to answer questions, but to help people choose paths with clarity and confidence.
Log in or sign up for Devpost to join the conversation.