Introduction

Overview

Asiyst is an enterprise-ready developer platform designed to embed interactive, visually grounded AI assistants directly onto modern websites and web applications. Unlike traditional text-only chatbots that sit passively in the corner of a screen, an Asiyst assistant pairs intelligent conversational models with animated visual avatars and real-time DOM interaction.

What is Asiyst?

Asiyst connects your website interface, business knowledge, and AI models into a cohesive, conversational guiding experience. The platform consists of four primary building blocks:

1. Web SDK (@asiyst/sdk): A lightweight client library that renders the visual assistant, synchronizes conversational state, and triggers client-side guidance actions.

1. Asiyst CLI (asiyst): A developer command-line tool for bootstrapping projects, scanning local repositories, syncing knowledge bases, and deploying assistant configurations.

1. Knowledge Engine: A multi-source indexing system that ingests websites, documentation files, GitHub repositories, and structured text to keep responses strictly factual.

1. Cloud Platform & Dashboard: The centralized control center for managing Projects, configuring Avatar Studio models, provisioning API Keys, and inspecting telemetry.

Core Architecture

An Asiyst deployment operates across secure browser, cloud, and edge boundaries:

┌─────────────────────────────────────────────────────────┐
│                      Your Website                       │
│  ┌───────────────────────────────────────────────────┐  │
│  │     @asiyst/sdk (Client Runtime & Avatar Widget)    │  │
│  └─────────────────────────┬─────────────────────────┘  │
└────────────────────────────┼────────────────────────────┘
                             │ HTTPS / WebSocket
                             ▼
┌─────────────────────────────────────────────────────────┐
│                  Asiyst Cloud Platform                  │
│  ┌─────────────────┐ ┌────────────────┐ ┌─────────────┐ │
│  │ Context Router  │ │ Knowledge Base │ │ Avatar State│ │
│  └────────┬────────┘ └───────┬────────┘ └─────────────┘ │
│           │                  │                          │
│           ▼                  ▼                          │
│  ┌────────────────────────────────────────────────────┐ │
│  │   LLM Inference (Claude / OpenAI / Custom Provider) │ │
│  └────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────┘

Key Capabilities

• Visual Grounding: Guide visitors with animated 3D and 2.5D avatars capable of idle breathing, walking transitions, and contextual gestures.

• Strictly Grounded Answers: Contextual retrieval from authorized documents and website pages eliminates hallucinated answers.

• Framework Agnostic: Integrates natively with Next.js, React, Vue, Svelte, or vanilla HTML script tags.

• Zero-Trust Security: Separate public SDK keys for browser runtimes from private management API tokens on your servers.

Next Steps

To deploy your first assistant in under five minutes, continue to the Quickstart Guide or explore SDK Installation.