Comfort comes first.
Familiar words. Small, useful steps. A quiet presence that makes space for your day.
A small, independent AI studio
Thoughtful AI companions for everyday life.
Made to feel familiar, fit your rhythm, and help with the things that matter to you.

Small companions.
Meaningful help.
We believe AI should feel approachable. So we're starting with the small moments: a thought worth keeping, a task made lighter, a little room to breathe.
01 / A companion taking shape
Early developmentYour everyday and your next big idea belong in the same life. We're shaping a personal AI assistant that helps you keep the thread between them.
Save a thought. Pick it up later. Find a useful next step.
I'd love a quiet weekend away. Somewhere green, with a good walking track.
That sounds lovely. Shall I keep this as a weekend idea for when you're ready?
Click through a scripted example. This is a design concept, not a live AI chat.
02 / A gentler kind of technology
We're designing for the feeling of being helped: comfortable, clear, and in your hands.
Familiar words. Small, useful steps. A quiet presence that makes space for your day.
Our goal is memory you can see, choose, and change. You decide what your companion carries forward.
Clear suggestions and a chance to check before meaningful actions. Helpful AI should leave the final say with you.
These principles guide what we're building, from the first prototype onward.
03 / From the studio
Stellmori is an independent AI software studio based in Auckland, New Zealand. We're exploring how personal AI can feel more useful, more approachable, and easier to welcome into daily life.
A background in backend engineering and system integration. A curiosity for making technology feel a little more human.
Mori's direction grows from Jarvis, an existing personal-assistant project. We're shaping that foundation into a focused product experience. Mori is in early development and isn't publicly available yet.
Our first focus is capturing a thought, returning to the right context, and turning it into a useful next step. Memory controls and clear confirmations are part of the design work.
We plan to evaluate Claude for understanding saved context, synthesising research, and planning useful next steps. This is a planned integration, not a claim of a live Claude-powered product.
A little curiosity goes a long way.