SpendingsBot

What it does

This assistant is capable of managing the user's personal expenses. The bot is able to :

- Record daily expenses.
- Restore information on the user's expenses.
- Create graphs from the user's expenses.
- Provide analysis and advice based on graphs.
- Create csv files of expenses based on user requests.

To perform these tasks, the bot uses several technologies:

1) Gemini: the bot uses several agents based on Gemini to perform the tasks listed above. Each agent has a specific role. These include :

- an agent that recognizes the user's intent.
- an agent that creates sql queries based on the user's request.
- an agent that creates the json to record an expense.
- an agent that creates code for graphics.
and more...


2) Google cloud APIs: Vision and Speech to text are the 2 Google APIs used for this bot. This allows users to send voice messages and images of their invoices directly to the bot. For example, when the user sends an image of an invoice, cloud vision performs the OCR and Gemini will create the json that will record the ticket information.

3) Big Query and cloud storage: The bot interacts with a BQ table, which groups all recorded expenses. It will also retrieve information from this table. When the user sends an image, in addition to recording the expenses on big query, the bot will also save the image in a cloud bucket to archive the invoice with a specific ID (which matches the relevant expense ID in BQ) .