Inspiration
What it does
How we built it
Challenges we ran into
Accomplishments that we're proud of
What we learned
What's next for GreenRoute Audit Lab
Inspiration
AI products often send every request to the same powerful model, even when a smaller model could meet the quality requirement. That can increase cost and estimated energy use while hiding where the request is processed. GreenRoute Audit Lab makes that routing decision inspectable before deployment.
What it does
The user describes an AI workload, chooses a required-quality setting and applies a data-region policy. A deterministic local policy engine classifies the workload, filters the demonstration model catalogue, selects the lightest adequate route and compares its estimated energy and CO2e range with a heavy-model baseline. Every decision can be exported as prompt-free JSON audit evidence.
How we built it
This is a new, standalone implementation created from an empty directory during the DevNetwork hackathon. It uses semantic HTML, responsive CSS and vanilla JavaScript. The prototype runs entirely in the browser: prompt text is neither transmitted nor stored. A versioned demonstration catalogue, explicit policy constraints and a deterministic audit schema keep decisions reproducible.
The founder's earlier GreenRoute MVP supplied domain learning only; this repository contains the new hackathon implementation and its own commit history.
Challenges we ran into
The hardest design problem was avoiding false precision. Provider hardware, batching, live grid mix and per-query electricity use are usually unknown. The interface therefore shows a ±40% range, labels results as decision-support estimates and never presents them as meter readings or verified avoided emissions.
A second challenge was treating quality and data location as real constraints. The router filters by region and capability before optimising estimated impact, so sustainability never silently overrides user requirements.
Accomplishments that we're proud of
- A working policy sandbox built from scratch during the event
- Transparent route explanations and rejected-alternative logic
- Responsive desktop and mobile interface
- Prompt-free, downloadable JSON audit records
- Public source code and reproducible setup with no paid infrastructure
What we learned
Efficiency claims are only useful when their assumptions, uncertainty and quality guardrails remain visible. A routing product needs an audit model, not just a model dropdown.
What's next for GreenRoute Audit Lab
Next steps are to replace illustrative coefficients with licensed or provider-published data, validate the classifier on an evaluation set, add live availability and grid signals, and expose the policy engine as an API or SDK for AI product teams.
Log in or sign up for Devpost to join the conversation.