Environmental Policy

Effective: January 2026

Class Architect recognises that the operation of AI-powered services carries a significant environmental footprint. We are committed to minimising our impact and operating in an environmentally responsible manner.

Our Commitment

We aim to achieve net-zero carbon emissions by 2030. This goal extends across all aspects of our operations, from model inference to office energy use.

Key Principles

• Energy Efficiency: We select AI inference providers that operate on renewable energy or purchase carbon offsets. Our model routing prioritises the most efficient model for each task, reducing unnecessary compute.
• Model Optimisation: We use quantised and distilled model variants wherever possible to reduce the energy cost per generation. We batch process spec uploads to maximise GPU utilisation and minimise idle energy waste.
• Infrastructure: Our cloud providers (Google Cloud) are selected in part for their commitment to carbon-neutral data centres and use of energy-efficient hardware.
• Paper Reduction: By enabling digital-first curriculum design, assessment, and feedback, Class Architect helps educators reduce paper consumption across their institutions.

Our Actions

• We aim to publish annual carbon footprint reporting
• Offset unavoidable emissions through verified carbon removal projects
• Encourage remote work to reduce commuting emissions across our team
• Review and optimise model inference efficiency quarterly
• Source hardware and cloud services from providers with strong environmental credentials

Measurement & Reporting

We track our energy consumption and associated carbon emissions on a quarterly basis. Our annual Environmental Report is published on this page. We welcome feedback from users and stakeholders on how we can further reduce our environmental impact.

Review

This policy is reviewed annually by our leadership team and updated to reflect new technologies, regulations, and best practices.

Class Architect • Aiming for net-zero by 2030