ByteScope

ByteScope shows how visible your business is to AI-asssistants, reveals how you're perceived, and gives clear tips to improve both visibility and sentiment

Comment

Inspiration

More and more people use AI assistants to plan their trips, to find a restaurant close to them or to decide on the next software for personal or professional use. While the focus still mostly remains on SEO, GEO is becoming more important by the minute. As I found that tools to check this were lacking, I wanted to build a fairly simple one that enables business owners to check their AI visibility and the perceived sentiment.

What it does

After answering a questionaire, that takes in all details about the business, a process orchestrated by several Supabase Edge Functions is started. First Gemini creates 10 questions based on the input and the analysis of the provided website. The questions are divided into three categories: Research, Information and Sentiment. The Research questions aim to check the recommendation of the business, the Information questions validate if the AI provides correct information and the Sentiment questions see how the AI judges the business.

These questions will then be sent to four different AI models (ChatGPT, Claude, Gemini, Deepseek) and will be answered just like any other users question would be answered. Finally Gemini analyzes the provided answers and generates an analysis out of them. These results are then displayed on a dashboard, with the opportunity to export the analysis details to csv for further analysis.

How we built it

The project is built entirely with bolt.new, while leveraging components from 21st.dev (Ready to use and Magic Chat). The backend functions are solely handled by Supabase edge functions. The database is also hosted in Supabase.

Challenges we ran into

I had a hard time setting up the orchestrated edge functions and also the prompting for the analysis took some time to get it right as the competitor analysis was not working at all times.

What we learned

The main learning I take from this hackathon experience is how to properly prompt. I learned the difference a great prompt can make.

What's next for ByteScope

ByteScope will become even better in the futures by adding fully automated retriggers of the analysis with results being sent straight to your inbox. This will happen together with the implementation of payments.

Built With

Share this project:

Updates