Usaid

Experience the future before you choose it

Comment

Inspiration

As humans, we are bad at long-term decision-making. We underestimate risks, overestimate our future motivation, and often ignore the second-order effects of our choices. Whether it's quitting a job, moving to a new country, or starting a company, we're forced to make life-altering decisions based on gut feelings and limited data.

We realized that while there are tools to track tasks and games to simulate fantasy worlds, there was nothing to simulate your own future. What if decision-making was like a video game? What if you could "save your game," try a risky path, see the consequences, and then rewind if it didn't work out?

This inspired Usaid, a cognitive time simulator that brings the power of "Scenario Planning" (used by military and corporate strategists) to personal life decisions.

What it does

Usaid is an AI-powered engine that generates multiple, distinct future timelines based on a single real-world decision.

  1. You enter a dilemma (e.g., "Should I quit my job to start a startup?").
  2. Usaid analyzes your profile (risk tolerance, values, current state) and generates few feasible divergent future paths (Optimistic, Pragmatic, Remote/Risky).
  3. Each timeline is visualized with year-by-year events and quantifiable metrics for Career, Finance, Relationships, and Mental Wellbeing.
  4. You can "inject" new decisions into any timeline (e.g., "What if I get funding in Year 2?") and watch the future rewrite itself in real-time.

It’s not just advice.. it’s experiential foresight of once own life.

How it is built

Usaid is a modern full-stack application leveraging the latest in Generative AI.

  1. The AI Engine: Google Gemini 3's advanced reasoning capabilities has been used to build the simulation core. Prompt Engineering is been used to build something beyond simple text generation by constructing a persistent "Cognitive State Model" for the user.

  2. The Frontend: Built with React, TypeScript, and Vite, featuring a premium "Glassmorphism" UI. Framer Motion is used for smooth timeline animations and Chart.js (or similar visualization libs) to make the data tangible.

  3. The Backend: A robust Node.js/Express server that acts as the orchestrator. It handles authentication, manages the SQLite database (via Prisma), and streams AI responses to the client.

  4. The Design: Heavy focus on "Industrial Polish", using a dark, futuristic aesthetic with neon accents to make the user feel like they are stepping into a control room for their life.

Challenges we ran into

  1. Hallucination Control: Getting an LLM to consistently generate plausible 5-year timelines without veering into fantasy was tough. Strict schema validation implementation and "Chain-of-Thought" prompting helps a lot.

  2. Latency: Simulating 5 concurrent futures is computationally expensive. Gemini 3 Flash know for its flash speed has been used and a streaming architecture has been implemented so users see the first timeline immediately while the others generate in the background.

  3. Visualizing Abstract Data: How do you visualize "Emotional Wellbeing" over 5 years? The UI has been iterated multiple times to find a graphical representation that was intuitive but not overwhelming.

Accomplishments that make me proud

  1. The "Wow" Factor: The moment you see your life split into three different paths on the screen is genuinely powerful.

  2. Structured AI: I managed to tame a creative LLM into outputting complex, mathematically consistent graph data.

  3. The UI: The interface feels incredibly premium and responsive. It doesn't feel like a hackathon project, it feels like a SaaS product.

What I learned

  1. Gemini 3 is a Reasoner, not just a Writer: If we give Gemini enough context about a user's psychology, it can predict second-order effects (e.g., "High career growth = Risk of burnout/divorce") with scary accuracy.

  2. The Power of Context: The quality of the simulation is directly proportional to how much the AI knows about the user. Personalization is the key.

What's next for Usaid

  1. Reality Feedback Loop: Allowing users to log what actually happened, so the system learns from reality and improves its future predictions.

  2. Multi-Agent Extensions: Adding specific AI advisors (a "Risk Manager," a "Career Coach," a "Therapist") who debate the pros and cons of your timelines.

  3. Scientific Extensions: Applying the same engine to simulate research paths or startup pivots for teams.

Built With

Share this project:

Updates