Energy Surge AI

• 

  Generate Images - Image shows levelized cost of energy.

• 

  Chat with Data

• 

  Create Plots

• 

Inspiration

Advancements in AI, industrial production, urbanization and economic development are driving the global demand for energy. According to the International Energy Agency, Global energy investments are projected to hit a record $3.3 trillion in 2025, with 2.2 trillion dollars going toward clean energy -renewables, nuclear, grids, storage. lEA

Despite this, navigating the energy landscape can be challenging, as energy data is often complex, involving large, multimodal, interdisciplinary datasets. Improving data accessibility and interpretation is essential for informed decision-making and for energy R&D.

What it does

Energy Surge AI is an energy data explorer agent that makes understanding complex energy data simple. Through interactions with the multi tool agent, users can intuitively explore data at multiple levels of resolution, including at the state, county level or by a specific geographic point coordinate using Natural Language. Users can generate solar potential summaries by county, query specific metrics such as capacity factor, total levelized cost of energy (LCOE), total levelized cost of transmission (LCOT) and solar installation developable area etc. The agent also allows for generation of plots and sample images that can be used for marketing and communications, or as suggestions for infographics in the preparation of reports.

How I built it

Energy Surge AI was developed using MongoDB and Google Cloud Integrations. Google’s Gemini text embedding models were used to generate vector embeddings, enabling semantic search with MongoDB Atlas Vector Search. The Google Agent Development Kit was used to build the agent with Gemini handling reasoning and language tasks and Imagen producing images from text prompts.

Challenges I ran into

Challenges in processing the datasets before uploading to MongoDB Atlas. Initially displaying images in ADK and adk deploy cloud_run.

Accomplishments that we're proud of

The agent is useful within the demo version for exploring the Open Energy Data Initiative datasets. The agent is scalable and can employ further integrations to enhance user interactions with other energy data sources (wind, geothermal etc.) It has broad applications for energy project developers, policy makers, urban planners and individuals etc.

What we learned

Google Agent Development Kit. MongoDB Atlas Vector Search.

What's next for Energy Surge AI

Further development of the Agent to include other data sources. Including more tools. Creation of custom UI.

Built With

• gemini
• google-cloud
• googleadk
• imagen
• mongodb
• pandas
• pymongo
• python
• vertexai