Rayanovin logo
AI & assistants

Rubah — a personal AI assistant on your own server

An AI assistant that runs on your own server and actually remembers you

AI assistants have become part of everyday work, but using the mainstream ones means sending your questions, files and internal company information to a server on the other side of the world — something many organizations simply cannot accept, on top of subscription costs, service outages and access restrictions. Rubah answers exactly that need: an AI assistant that runs entirely on your own server, sends nothing outside, and costs nothing per message. Unlike ordinary chatbots Rubah has memory; it remembers what you have told it about yourself, your work and your preferences, so you never have to explain everything from scratch again. Each user can give the assistant its own name, tone and personality, and refine it over time simply by approving or correcting its answers. For companies that means an internal assistant that answers repetitive staff questions, helps draft and summarize text, and keeps organizational knowledge inside the organization. For individuals it means a permanent companion on a home server. Rubah can also live inside a desk device or small robot, giving hardware makers a ready foundation for adding a talking assistant to their product.

Python 3.12FastAPISQLAlchemyPostgreSQLQdrantRedisarqOllamaGo 1.23WebSocketDockerCaddy
Screenshots for this product are coming soon
Key capabilities

Everything an organization needs

Runs on your own server

No question, file or piece of information leaves your organization or home — everything is processed on your own hardware.

Persistent personal memory

Rubah remembers what you've told it about yourself, your work and your preferences, so conversations pick up where they left off.

A personality per user

You choose the assistant's name, tone and character, and every user gets their own version.

Natural conversation

Built on a language model tuned for the local language, so answers read naturally rather than machine-translated.

Learns from your feedback

Approving or correcting answers gradually makes the assistant more accurate and better matched to how you work.

No per-message cost

Unlike cloud services there's no foreign-currency subscription and no charge per message — deploy once, use always.

Dedicated hardware or plain desktop

It can live inside a robot or a desk device and talk to you, or simply run on a computer or server.

Optional help from larger models

For complex questions it can optionally call out to cloud services — a setting you can leave switched off entirely.

What's inside

The parts that make up the platform

Rubah (RNT.Fox) follows enterprise architecture principles: separated services, room to scale and full monitoring from day one.

  • Local language model on your own server: the core of every answer
  • Per-user memory: securely stored facts, interests and events
  • Personality builder: name, tone and behaviour per user
  • Feedback loop: gradual improvement from approvals and corrections
  • Conversation client: an app or dedicated hardware to talk to the assistant
  • Security layer: complete isolation of each user's data
Technical specifications

Specification table

Backend
FastAPI / Python 3.12 (async)
Database
PostgreSQL 16
Memory search
Vector database, isolated per user
Queue
Redis
Language model
Local, Persian-tuned
Robot client
Go binary
Deployment
Docker Compose + HTTPS
Data location
Entirely on your own server
Use cases

Which organizations it fits

Companies & office teams

An internal assistant for repetitive staff questions, drafting and summarizing

Sensitive & government organizations

Using AI without sending a single byte outside the organization

Support & sales teams

Helping agents answer faster and more consistently

Individuals & home users

A permanent assistant with personal memory on a home server

Hardware & smart product makers

Adding a talking, local-language companion to a device

Research & academic groups

A ready platform for working with language models on internal infrastructure

Ready when you are

Let's build your organization's next platform together.

A free 30-minute discovery session to review your needs, pick the right product and lay out a clear execution path.