What it is
An AI chat organised around the parts of your life. Everyday, Work, Health, Life admin and Relationships each have their own space, with their own instructions and the model that suits them.
Who it's for
People who already use AI and want to use it with more care: no hype, no guilt, and no bigger model than the question needs.
What it helps you do
Dinner from three ingredients. The message you don't know how to answer. A job application, a sleep question, a trip with the kids.
Ask, get an answer, see which model gave it
I've got eggs, spinach and half a lemon. What's for dinner?
A spinach frittata. Whisk 4 eggs with salt and pepper, wilt the spinach in a little oil, pour the eggs over, cook on low for 6 minutes, then 2 minutes under the grill. Squeeze the lemon over it at the table. Twenty minutes, one pan.
Turn these notes into a short brief: new onboarding flow, three screens instead of eleven, ship by Friday, Rika owns the copy.
Project brief: onboarding in three screens
Goal: a new account reaches its first question on the third screen.
Scope: name, one question about where they heard of us, plan choice.
Owner: Rika (copy), Fredrik (build).
Deadline: Friday.
Out of scope: the landing page and emails.
Two questions, two different models. The first needed a light one; the second needed a stronger one. Every reply tells you which.
Four things we do differently
The smallest model that does the job
Your question goes to one of four providers: Mistral (France), GreenPT (Netherlands, open models on renewable-powered servers), OpenAI and Anthropic. A light model answers first, and a question that needs more steps up. Every reply says which model answered.
Spaces, not one endless chat
Each part of life gets its own space with its own instructions, so your work notes and your sleep questions never share a thread.
Eco mode
Shorter answers for the quick questions. Flip it on per chat, and the assistant keeps it brief.
AI's energy cost, honestly
AI has a real cost for nature. We can't make it zero, so we use as little as each question needs, and we don't offset or call anything neutral. Research from UCL and UNESCO shows small, task-specific models can use 15 to 50 times less energy than a large general one⁷.
Your chats are yours
Our team does not read your conversations, we don't train AI on them, and ads on the free plan are never based on what you write.
We will fund community-led ocean restoration projects
Each paid plan will fund one mangrove tree a month once our restoration partnership is signed. We'll name the partner here when the agreement is.
Bring your people. More of us means more restoration we can fund.
Technology should give back more than it takes
AI adoption isn't slowing down. The relevant question is what that growth costs the planet, and what we do about it. We built We love blue to use as little AI as each question needs, and to be open about the rest.
We love blue started from an observation: people now use AI for everything. They organise their exercise, search for recipes, and create bedtime stories for their kids. Meanwhile, the major AI companies keep racing to build better models. But a better model isn't the whole answer. How people actually work with AI, and what that work costs the planet, matters just as much.