Last year, when the Trinidad and Tobago National Budget was read in parliament, I had a wild idea. What if we could use AI to generate intelligent summaries of the budget tailored to our interests? Or better yet, what if we could ask general questions and get a breakdown of spending?

At the time, I wasn’t an AI developer, and I didn’t know where to start. But I had a hunch that there was a way to make this happen.

I’d been using a machine learning-based approach to review my personal budget, which worked well. But the national budget document was over 100 pages of dense text, making it too complex for me to tackle alone. So, I put the idea on hold, hoping that one day I’d find a way to make it happen.

And then, out of the blue, I came across a tweet that changed everything. @mayowaoshin described how he used GPT-4, LangChainAI, and Pinecone to “talk” to a massive legal document – a ruling. I knew right away that this was the answer to my problem!

Mayo shared his work on GitHub, and within 30 minutes, I’d applied his method to my scenario. The result? I could finally “talk” to my national budget using the mechanics of chatting with GPT. Check out the video below to see it in action:

When I first had the idea of using AI to understand the national budget, I didn’t even know about ChatGPT. But the incredible work that OpenAI has done with GPT, Dall-E, and other applications is making it possible for us to achieve things we never thought were possible.

Using AI to understand complex documents like the national budget has enormous potential. With this technology, we can gain insights and make informed decisions. The possibilities are endless, and I can’t wait to see what the future holds for AI and its impact on our world.