Inspiration
With the rapid rise of AI, large language models (LLMs) are becoming a core component of our digital infrastructure. However, most AI prompts are inefficient, wasting compute. Every extra token increases cloud cost, latency, and energy consumption in data centers.
This was inspired by a simple question: What if we could make every AI query cheaper, faster, and greener automatically?
How we built it
Flutter for the interactive UI Google Gemini API for semantic compression and verification Firebase Firestore to log global token savings in real time
Challenges we ran into
Preventing the AI from answering prompts instead of compressing them Designing metrics that meaningfully connect token reduction to cost and energy impact Avoiding wasted API calls for already-efficient prompts
Accomplishments that we're proud of
Built a real-time semantic compression pipeline with fidelity verification Reduced prompt size without losing meaning Created a global savings dashboard using Firestore Designed a system prompt that resists prompt injection and command execution Quantified environmental and economic impact of prompt optimization
What we learned
Most AI prompts contain large amounts of redundant language Prompt optimization can significantly reduce cloud inference cost and latency
What's next for Eco Prompt
Browser extension to automatically compress prompts in ChatGPT, Gemini, and other AI tools Support for structured formats (code, JSON, YAML, research papers) Public leaderboard
