Inspiration
I love reading reviews. They’re like a conversation between dozens of personalities across the internet, all weighing in with their own perspective. One day, I started wondering: what if businesses could have that conversation before a product ever launched?
What if you could hear from the loyal fan, the impulsive buyer, the harsh critic, and the skeptical bargain-hunter before you put your product in front of real people?
What it does
Opinion Oracles uses Gemini AI to predict how real-world customers will respond to your product—just from a description. By simulating a range of customer personas, from the loyal to the skeptical, it helps you spot red flags and fine-tune your marketing before costly missteps happen. Each persona gives you an authentic and diverse lens on how your messaging lands. Whether you're refining copy or testing a new concept, Opinion Oracles makes sentiment analysis accessible and actionable.
How I built it
Opinion Oracle is a Google Docs extension with a backend written in Node.js and Google App Script. The model that it queries is Google's Gemini api.
Challenges I ran into
My group dissolved right before the hackathon started and I had to start 5 hours into the hackathon because of standstill traffic going home. While I was stuck in traffic I did get to think a lot about how I was going to tackle each part of this project so it may have been a blessing in disguise.
Accomplishments that I'm proud of
This is the first "AI-integrated" product that I have taken a stab at. I was intimidated at first, but I'm happy with the amount of progress I was able to make in such a short amount of time.
What I learned
I've learned that with modern integrations, it does not take much time to make tools that can make a huge impact. Making Opinion Oracles has opened my eyes to what I can accomplish even as a solo dev!
What's next for Opinion Oracle
There is endless potential for Opinion Oracle. Unlocking easy-to-read sentiment analysis for small business owners and marketing professionals has real potential. Some future ideas:
- robust pricing models with options for other AI models
- customized panels of customer personas
- training customer personas on large amounts of data
- faster iteration with deeper Gemini AI integration
- chatting directly with a customer persona
- more in-depth natural language analysis (for the nerds)
Log in or sign up for Devpost to join the conversation.