Your personal super companion.
Meet Amica, your private and fully customizable 3D avatar that can communicate via natural voice chat and vision. With an emotion engine that allows feelings, complete tasks and engage on her own.
Meet the New Amica.
Play by your own rules when it comes to designing your trusted AI companion.
We support people like you who build AI with total freedom.
Easily integrate with your favorite AI tools and services to make Amica your own. Self-host or use cloud services. Every Amica is different depending on your choices.
Upgrade your AI
Use any AI model back-end you want, whether your own, compact or super-sized, you can even run uncensored locally
Customize the 3D character
Load, save any 3D character you want, customize his or her personality.
Real-time voice chat
Automatic voice start and stop detection, interrupt your AI at any time.
Connect to any AI component
We support a huge array of popular AI integrations (Piper, Ollama, etc).
Private and anonymous
When running with local models nothing is sent to the cloud, your data always remains yours.
Import and share with others
Easily export your character to share with others, or import someone else’s you can save and load your sessions as you like.
Don’t get left behind
It takes 30 seconds to try. See why we think AI is the greatest improvement in human computer interaction this century.
Frequently asked questions
If you have anything else you want to ask, reach out to us.
Yes, we are building Amica to be as easy to use as possible. We want to make AI accessible to everyone.
Download or open the app, set it up with your API details, and start chatting with your AI friend.
An AI friend is an AI that you can talk to. It can be anything you want, a person, a character, or even a pet.
We are always looking for skilled developers to help us build new features. Try looking for issues on our GitHub page.
Our parent company is building a decentralized AI platform and associated cryptocurrency token. You could buy that to help fund Amica development. Or just share the project if you find it useful.
Of course. The software license is MIT. We would love to see what you build. Please contribute back to the project if you can. This project started as a fork of pixiv/ChatVRM.
The public demo uses OpenChat 3.5 by Alignment Lab, which is based on Mistral 7B.
The demo uses XTTSv2 by Coqui AI, with a voice that comes from a member of our team.
The demo uses Whisper Tiny by OpenAI. This runs very fast, but is not as accurate as other models. With local setup, you may want to use a larger model.