Build Agent App With LlamaIndex TypeScript
Learn how to build, deploy, and integrate agent applications with comprehensive tools and multi-platform support. Quick StartQuick Links
Build
Create agents with your chosen framework
Deploy
Deploy to serverless platforms, cloud infrastructure, or self-hosted
Integrate
Build agent applications with chat UI, generative UI, human-in-the-loop workflows, and client SDKs
How to Build an Agent
Building an agent involves defining agent logic, connecting to LLMs, and integrating tools. The framework provides a cohesive system for all agent development needs.| Goal | What to use | Description |
|---|---|---|
| Get started quickly | Quick Start | Build your first Agent in 10 minutes with step-by-step guide. |
| Create agent logic | Core Concepts | Architecture for creating agent workflows. Brings models, tools, state, and logic together. |
| Connect to LLMs | Model Providers | Support for multiple LLM providers with unified interface. |
| Add external capabilities | Tools | Access to external services with custom tools and function calling. |
| Manage conversation state | Memory & State | Session-scoped state management and conversation history for contextual responses. |
| Run your agent | Run Agent | Start your agent service and connect it to client applications. |
Deploy Agents in Your Product
AG-Kit provides flexible deployment options to integrate agents into your existing infrastructure or create new applications.| Goal | What to use | Description |
|---|---|---|
| Deploy to cloud | Cloud Deployment | Deploy to serverless platforms, Docker, or Kubernetes. |
| Self-host deployment | Self-Hosted | Deploy on your own infrastructure with full control. |
| Monitor performance | Observability | Built-in monitoring and observability features for production deployments. |
Build Agent Applications
Create comprehensive agent applications with advanced UI components and user interaction features.| Goal | What to use | Description |
|---|---|---|
| Create chat interfaces | Chat UI | Create conversational interfaces with chat components and hooks. |
| Generate dynamic UI | Generative UI | Build dynamic, interactive user interfaces based on tool results. |
| Add human interaction | Human-in-the-Loop | Implement interactive workflows requiring human input and approval. |
| Create client side tools | Client Side Tools | Define custom UI actions and interactions for agent workflows. |
| Manage shared state | Shared State | Handle state management across multiple components and sessions. |
Key Features
Multi-Platform Deployment
- Web Applications: Components with streaming support
- WeChat Mini Programs: Native integration with AG-UI protocol
- Desktop Apps: Support for cross-platform desktop applications
- Serverless: Optimized for serverless platforms
Advanced Capabilities
- Generative UI: Dynamic UI generation based on tool results
- Human-in-the-Loop: Interactive workflows requiring human input
- Multi-Agent Systems: Coordinate multiple agents for complex tasks
- Memory & State: Persistent conversation memory and session management
Developer Tools
- CLI Tools: Command-line interface for project management and deployment
- Hot Reload: Fast development iteration with automatic reloading
- Type Safety: Full type support with comprehensive type definitions
- Testing Suite: Built-in testing framework for all agent components