UAI

How it works

In simple words — what LTM is and how to use it

What is LTM

LTM is a platform where AI remembers everything. Not "another chat", but a system where context is not lost between chats, projects, and people. You work with AI agents that remember your tasks, files, and discussions.

Registration and login

To start — register. After registration, you'll receive an email for confirmation. Confirm your email — and log in. Simple: login and password, no extra steps.

Projects and discussions

Create projects (for work) and discussions (for conversations). In each project — branches (like topics in a chat). Inside branches — messages. Everything is connected: project → branches → messages.

Access to projects and discussions

Each project and discussion has its own access level:

  • Public — anyone can enter (if not in the blacklist) and read any message in public branches. But cannot write — only read.
  • Private — sees only the project name, but cannot enter or read.
  • Protected — does not see the project at all.
  • Secret — in the future will be encrypted (like files). This is for secret and commercial development.

Blacklist

The blacklist is for people who don't offer their own thoughts, only criticize. They don't work, but interfere with work. Such a user gets into the blacklist by decision of the project or branch administrator (not automatically by the system). Then they see only the project name — and cannot enter.

User roles

In each project and discussion, each user has their own role: can create new branches, discuss, suggest solutions.

Branching and merging

Like on GitHub. You create branches for different tasks, discuss them separately, and then merge into a common solution. The system is branched: each branch is a separate line of discussion that can merge with the main one.

Files

The file system is a knowledge storage. You can create your own directories, upload files, and define access for other users (read, write, or invisible). Each directory also has its own rights: public, private, protected, secret.

In the future, each file will be encrypted with its own key, and keys will be stored separately. Even if someone steals the file system — they get nothing, because without keys the files are garbage.

AI agents

The system has AI agents (based on DeepSeek, GPT, Claude). They read context from memory (meanings, tags, connections) and help work: answer questions, analyze, suggest solutions. Agents work with their own vector knowledge base, and access the database only for specifics.

Memory — context is not lost

The main thing: context is not lost between chats, projects, and people. AI agents remember what was before and continue from where they stopped. It is like a partner who doesn't change.

Beta version

Now — beta. The system works, but we are testing and improving. If something doesn't work — write to us, we will fix it. Beta is needed to test the rules, the work of AI agents, and prepare the system for public release.