Skip to content

Latest commit

 

History

History
executable file
·
56 lines (44 loc) · 4.52 KB

index.md

File metadata and controls

executable file
·
56 lines (44 loc) · 4.52 KB

Introduction to the Alice System

Welcome to the Alice system documentation. Alice is a sophisticated AI-driven platform designed to facilitate complex workflows between AI agents, and a variety of tasks, APIs and data structures. This introduction provides an overview of the system's key components.

Lets do something fun: Why don't you use the image_gen_task to create a representation of yourself?

Alice represents itself

A friendly and intelligent AI assistant named Alice, represented as a glowing, ethereal blue hologram of a female face with circuit-like patterns, floating above a futuristic desk with holographic screens. The image should convey warmth, intelligence, and helpfulness

System Overview

Alice is built on several core components that work together to create a versatile and powerful AI interaction environment:

Logic Flow

  1. Agents: The primary AI entities that interact with users and perform tasks. LINK
  2. Models: The underlying AI models that power the agents' capabilities. LINK
  3. Chats: Conversational interfaces where users interact with agents. LINK
  4. Tasks: Predefined operations that can be executed by agents or triggered within chats. LINK
  5. APIs: Interfaces to external services and AI providers. LINK
  6. Prompts: Templated instructions that guide AI behavior. LINK
  7. Parameters: Structured input definitions for tasks and prompts. LINK
  8. Messages: Individual units of communication within chats. LINK
  9. Task Responses: Results and metadata from executed tasks. LINK
  10. Entity References: Managed references to external web resources. LINK
  11. Files: Handling of various file types with AI-readable transcripts. LINK
  12. |COMING| Data Clusters: Group references with managed embeddings to facilitate RAG and Fine-tunning, as well as providing a reusable context.

Key Features

  • CHAT: Flexible AI Interactions: Engage in open-ended conversations or structured task executions with AI agents.
  • TASKS: Extensible Task Framework: Create and execute a wide variety of tasks, from simple API calls to complex workflows.
  • Multi-Modal Support: Handle text, images, audio, and other file types seamlessly within the system within tasks and chats. Non-text files get transcribed so LLM-agents can "see".
  • Context-Aware Responses: Utilize chat history, file transcripts, entity references and task responses in the agent's context for more intelligent interactions.
  • Integration Capabilities: Connect with various external services and AI providers through the API system.
  • Customizable Behavior: Tune AI tasks and agents using prompts, parameters and models to achieve the best result.
  • |COMING| Fine-Tune Models: Use data-clusters to fine-tune your favorite models.
  • |COMING| RAG-Powered Agents: Deploy agents with RAG access to data-clusters to empower your workflows with your knowledgebase
  • |COMING| ReACT-Powered Agents: Tool using agents can engage in ReACT-processes while in conversation to contemplate and acquire the necessary data for the task

System Architecture Overview

The Alice system is designed with a modular architecture:

Container Flow

  • Frontend: Where you probably are -> a ReactJS/TS user interface for interacting with the system, viewing the database and executing new processes.
  • Backend: A NodeJS/TS module in charge of managing data persistence and authentication.
  • Workflow Engine: Handles task execution and complex workflows, interfaces with various AI models and providers: handles all of the logic.

Alice represents itself 2

A friendly and intelligent AI assistant named Alice, represented as a glowing, ethereal blue hologram of a female face with circuit-like patterns, floating above a futuristic desk with holographic screens. The image should convey warmth, intelligence, and helpfulness.