Core Concepts
What is Asiyst?
Deep dive into the Asiyst platform philosophy, visual agent paradigm, and architecture.
Overview
Asiyst is an interactive AI assistant platform built specifically for modern digital products and web applications. Traditional website chatbots rely on static text boxes isolated from the host page. Asiyst redefines this interaction by introducing Visual AI Agents that actively understand user location, guide users across DOM elements, and speak with grounded organizational intelligence.
The Visual Assistant Paradigm
Visual guidance creates higher engagement and faster comprehension:
- Visual Presence: Visitors interact with an animated 3D or 2.5D character that reacts with conversational gestures, walking cues, and idle animations.
- Context-Aware Assistance: The assistant detects current URL parameters, active page routes, and user interactions to offer timely assistance without intrusive popups.
- Grounded Responses: Powered by vector embeddings generated directly from your verified documentation, files, and repository sources.
How Asiyst Differs from Traditional Chatbots
| Feature | Traditional Chatbots | Asiyst Visual Assistants |
|---|---|---|
| Interface | Isolated text box | Interactive animated avatar with floating bubbles |
| Grounded Intelligence | Generic system prompts | Strict multi-source RAG with source citations |
| Developer Ergonomics | Proprietary iframe embeds | First-class Web SDK, CLI sync, and REST API |
| Action Execution | Text answers only | Page navigation, scroll targeting, and element highlighting |
The Four Core Layers
- Presentation Layer: The Web SDK managing Canvas / WebGL rendering, conversational speech bubbles, and sound effects.
- Orchestration Layer: Manages session state, conversation persistence, rate limiting, and domain validation.
- Intelligence Layer: Combines LLM inference (AI Providers) with dense vector retrieval (Knowledge).
- Management Layer: The developer dashboard, Asiyst CLI, and REST API for continuous delivery.