ARIA - Personal Research & Decision Making Assistant

An AI personal research & decision making assistant that helps you make the right personal, career, business, and investment decisions really fast by turning internet data into poll simulations.

Comment

Inspiration

I love market research as a way of making the right decisions. I have worked in market research and love looking at poll/survey data to make better sense of the world. Through my work, I saw how transformative quality market research can be for decision-making in large companies—but also how inaccessible it remains for individuals and small businesses. Traditional research costs thousands of dollars and takes weeks/months - to conduct surveys/interviews and decipher the data. Meanwhile, the internet is filled with data buried in research reports, reviews, discussions, blogs, etc. So the question is - how do you collate data from different sources and present them in a way so that you & I can make quality decisions in seconds?

What it does

Think of Aria as your personal research team that instantly analyzes hundreds of sources, simulates what experts and real people would say, and shows you the smartest choice through interactive polls & infographics. It's Perplexity meets Pinterest.

Core Capabilities:

  1. Natural Language Queries: Ask any decision question in plain English (e.g., "Should I start a sustainable fashion brand targeting Gen Z?")
  2. Synthetic Survey Generation: Creates simulated polls with representative demographic breakdowns without collecting new data
  3. Multi-Source Intelligence: Aggregates signals from reviews, forums, social media, search trends, job markets, and public datasets
  4. Save Decisions as JPGs

How we built it

I already had the idea in my mind and had a rough sketch of what the ui should look like. So I built a React app using gemini-3-pro-preview with googleSearch grounding, integrated with Firebase Firestore for cloud persistence and Recharts for data-driven visualizations.

Challenges we ran into

  1. The right UI to help a user make decisions in seconds. Still improving on it.
  2. Citing more reliable sources of data. Working on it.
  3. Improving on prompting Gemini to get better outcomes

Accomplishments that we're proud of

  1. Compressed complex research into scannable visual formats (polls, confidence bars, action plans)
  2. 30 seconds to get research that would take hours manually
  3. I & a few users are using the app for decision making

What we learned

  1. How to use firestore database to store the research data
  2. Already showcased the app on a few Reddit communities with positive response. The current consensus is ARIA is like Perplexity with cool UI.
  3. Using Gemini's thinkingBudget is critical for high-stakes decisions, as it allows the model to reconcile conflicting search data before presenting a final simulation.

What's next for ARIA - Personal Market Research Assistant

The next steps are:

  1. add user auth
  2. add subscriptions
  3. Improve on the challenges mentioned including making the AI more confident about the decisions and adding more reliable sources of data

Built With

Share this project:

Updates