Project Alice is an OS agentic workflow framework that integrates task execution and intelligent chat capabilities. It provides a flexible environment for creating, managing, and deploying AI agents for various purposes, leveraging a microservices architecture with MongoDB for data persistence.

For a live demo, visit here.

To see the repository, visit here.

The project consists of three main components:

Backend (Node.js with Express - TS) -> Manages the MongoDB, LM Studio generations and the file system, including serving files for the frontend.

Workflow (Python - Pydantic) -> Handles (most) of the logic, interacts with external APIs, consumes the Database through the Backend, and reads from the file system. Main endpoints are task execution and chat response generation.

3. Frontend (React - TS) -> UI that consumes/interacts with the DB and file system through the Backend and calls Workflow's endpoints to trigger executions.

The framework is based around 4 main components:

APIs and their engine

Agents, which deploy prompts and have models for any API they want to use

Tasks that leverage agents, other tasks and APIs to produce an output

Chats, that leverage tasks and agents, to produce a conversational experience

These components share information in one of 6 main ways, all of which have a string representation:

Files (All file types have a method for generating a 'transcript' for the file, and files generated through prompts keep it as a representation)

Messages

Task Results

Entity References

User Interactions

Embeddings