Moh, the Business Analyst Chatbot

Inspiration

Working with ChatGPT and Neo4j is what inspired me to make this application, I My interest in wanting to build something with Chat GPT and Neo4j is what inspired me to create Moh. So when I saw that Square and Google were teaming up to sponsor a hackathon I saw the perfect opportunity. Since the data available via the Square api is highly connected and it provides a standard data structure it made a perfect candidate for an application like Moh.

What it does

Moh is a business analyst chatbot that can answer questions about a business's data that is collected on Square. Users can interact with Moh by texting their questions and it'll respond with an answer.

How we built it

Moh was built using Dialog Flow ES and Twilio for the user interface, Chat GPT api to convert natural language to a query and create a human-like response to the question. To store the data I used Neo4j and to source the data I used the Square api.

Challenges we ran into

Challenges included getting sample data to test the app, integrating all the technologies, and balancing my time between family, work and this project.

Accomplishments that we're proud of

Being able to translate questions from the users to a db query and create a human-like response. Fine-tuning my first Chat GPT model and finally being able to build an app using a graph database.

What we learned

I learned more about Google Cloud, particularly Dialog Flow, Cloud Functions, and Cloud Run. How to work with and fine-tune a Chat GPT model to pull data from a graph database

What's next for Moh, the Business Analyst Chatbot

Next steps for Moh are to continue fine-tuning the model to improve the response quality, adding visuals along with the response, and increasing its analytical power by enabling it to perform ML tasks.

Built With

• chat-gpt
• google-cloud
• neo4j
• openai
• python
• square
• twilio