Carbon Optimised Search Engine: WillowAI

AI runs on servers, so let’s make them smarter and greener. Our search engine uses AI to route queries through low-carbon servers, optimizing results while cutting emissions. Smarter AI, lower impact.

Comment

Carbon-Optimized Search Engine: Reducing AI's Environmental Impact

🚀 Inspiration

The rise of AI-powered search engines and large language models (LLMs) has led to massive energy consumption, with data centers consuming nearly 1% of global electricity. Search queries are often processed based on proximity rather than sustainability, leading to high-carbon emissions from fossil-fuel-dependent servers. Our team was inspired to create a solution that optimizes query routing using real-time carbon intensity data, ensuring searches are handled by the most energy-efficient and low-emission infrastructure available. Our approach aligns with growing regulatory pressures, such as EU Digital Sustainability and Net Zero Goals, while providing a seamless AI-powered search experience.

🔍 What Does Our Solution Do?

Our Carbon-Optimized Search Engine, WillowAI, intelligently routes search queries to data centers operating on the cleanest available energy sources in real time. By leveraging AI and carbon intensity tracking, our search engine:

✅ Detects query location & energy grid status
✅ Routes searches to low-carbon servers for AI processing
✅ Optimizes LLM compute loads based on grid sustainability
✅ Provides users with insights into the carbon impact of their searches

By dynamically adjusting where and how queries are processed, we reduce AI-driven carbon emissions without compromising search speed or relevance. Additionally, our system aligns with corporate ESG (Environmental, Social, and Governance) goals, offering organizations a transparent way to reduce their digital carbon footprint.

🛠 How We Built It

Our team followed an agile development process, rapidly prototyping and validating key functionalities. We used:

  • Frontend: HTML, CSS, JavaScript, React.js (with Tailwind CSS) for a dynamic, user-friendly search interface.
  • Search Engine: Elasticsearch for indexing and retrieving results efficiently.
  • Carbon Tracking: Integration with ElectricityMap API and WattTime API to fetch real-time carbon intensity data.
  • Machine Learning: Python (Scikit-learn, PyTorch) to optimize routing decisions and compute load balancing.
  • Database: PostgreSQL for structured query logs and sustainability tracking.
  • Deployment: Vercel for frontend, Railway for backend, and Docker for containerization.

📊 Validation Testing:

  • Conducted stress tests to evaluate search speed vs. routing efficiency.
  • Benchmarked carbon reduction per search against conventional search engines.
  • Simulated various grid conditions to optimize real-time adjustments.

⚡ Challenges We Faced

  • Sourcing real-time carbon emissions data for global data centers proved complex. We resolved this by establishing we would use Climatiq API.
  • Programming without API key due to paywalls made it difficult to develop the backend to ensure it effectively worked.
  • Balancing sustainability with speed required extensive testing and refinement of our AI-driven query routing model.
  • Ensuring seamless user experience while dynamically routing search queries without noticeable delays.
  • Adapting AI workloads dynamically to reduce carbon impact while maintaining accuracy and relevance.

🏆 Accomplishments & Learnings

✅ Successfully implemented real-time carbon-aware query routing.
✅ Achieved measurable reductions in search-related carbon emissions.
✅ Validated that AI-powered searches can be optimized for sustainability without loss of accuracy.
✅ Gained deep insights into the intersection of AI, sustainability, and infrastructure efficiency.

🚀 What’s Next?

Short-Term Improvements:

  • Further optimize query routing algorithms for faster response times.
  • Improve user-facing sustainability insights, such as carbon savings per search.
  • Enhance ML-based adaptive compute scaling for better energy efficiency.

Long-Term Scalability:

  • Expand data partnerships with cloud providers to directly integrate with their sustainability metrics.
  • Develop an API service allowing other AI applications to leverage carbon-aware compute decisions and reduce the need for the Climatiq subscription.
  • Work with regulatory bodies to support global adoption of sustainable AI infrastructure policies.

With more time and resources, we envision a fully scalable, enterprise-ready AI search platform that makes AI-powered queries smarter, faster, and greener.

🌍 Join us in revolutionizing AI search for a low-carbon future! 🚀

Share this project:

Updates