Reference · Environment
AI and the environment.
AI products use real compute, which uses real energy. Here's how Continio thinks about that, and what it actually does about it.
The honest position
Continio is not zero-impact. No AI product is. Every message sent, every response generated, every background task consumes compute. We're not going to pretend otherwise.
What we can say honestly is this: Continio is designed to use less compute per useful output than the alternative, and we make deliberate technical choices that reduce energy use per message.
How Continio reduces compute per message
The biggest source of wasted AI compute is re-explaining context. Every time a user starts a new chat and re-introduces themselves, their situation, their project — that's compute spent re-processing information the system should already know. Continio's core purpose is to eliminate this. Fewer redundant tokens means less energy per useful response.
Continio uses prompt caching for system prompts (the instructions sent to the model on every message). A cached prompt costs a fraction of the energy of a full inference. On average, this saves 90% of the compute cost of the system prompt across repeated calls within a session.
Not every message needs the largest, most capable model. Simple questions go to lightweight models (GPT-4o mini, Claude Haiku). Complex, nuanced, or personal conversations get the full model. Routing smaller tasks to smaller models is a direct reduction in per-message compute — and it makes responses faster too.
Background enrichment tasks (anchor extraction, embeddings, summaries) pause when you're actively chatting. They run during idle periods. This avoids unnecessary concurrent compute load and keeps the system responsive when you need it.
Infrastructure
Continio runs on Vercel (frontend), Railway (backend), and Supabase (database and storage). AI inference runs on Anthropic and OpenAI infrastructure. All three major cloud providers have published net-zero and renewable energy commitments:
- Google Cloud (Anthropic) — sustainability.google
- Microsoft Azure (OpenAI) — microsoft.com/sustainability
We don't claim credit for their commitments — but it's worth knowing that the underlying infrastructure is not running on coal.
What we don't claim
We're not carbon neutral. We don't have a sustainability certificate. We haven't calculated our emissions per user per month. These would require more data than we currently have, and publishing a number we can't verify would be worse than saying nothing.
What we do commit to: as the product grows, efficiency improvements come before feature additions. Routing lighter tasks to lighter models, caching more aggressively, eliminating redundant processing — these are engineering priorities, not afterthoughts.
The comparison that matters
The relevant question isn't "is this zero impact?" — it's "is this better or worse than the alternative workflow?" A user who repeatedly re-explains their context across five different AI tools, restarts conversations, re-uploads files, and re-processes the same background information multiple times per week is generating more compute load than the same user with persistent, well-routed continuity.
Continio's efficiency argument is structural, not incremental. The product is designed around the principle that you should have to explain things to your AI tools once, not repeatedly.