GreenGeek

• 

  Carbon Emission Overview

• 

  Carbon Emission Prediction

Imagine knowing the climate impact of every purchase you make -- your coffee, your groceries, even your Netflix subscription. It would be empowering right? It would change how we think about consumption and help in sustainable future. So, we came up with this idea of how customers of bunq can be empowered to share in building a sustainable future.

We started with Agents based approach combining with Gemini and a wrapper around python-sdk. The agents take care of fetching the user's transaction history, categories them based and uses a hybrid matching scheme to calculate the carbon footprint for each category. To motivate customer's to reduce their carbon emissions by suggesting them some percentage of their expense to balance the carbon offset. Then, if the customer agrees the money is transferred to a green Bunq account all this through agents.

Next, we went one step forward and put all these tools in different MCP servers, so that the agents decide best tool based on the use-case of user. We integrated our local MCP servers with the claude app to do the same process seamlessly.

We also show some insights for the user on a dashboard and also worked on creating a UI interface to move forward from cloud and use some other LLM, but we could not finish it completely on time, still we added some screenshots of that as well.

PS: Our Demo through Claude with better quality is at the end of video, our AI skills are much better:D

Built With

• agents
• gemini
• llm
• mcp
• pydantic-ai
• python
• streamlit