Skip to main content

​
What is an Application?

An Application is an AI agent instance that lives inside a Workspace. Each application combines a model configuration, optional knowledge bases, tools, and conversation settings into a deployable unit that can serve users through chat, APIs, or external messaging channels.

​
Application Types

Nadoo AI supports three application types, each designed for a different complexity level.

Chat App

A conversational chatbot powered by a selected AI model. Attach knowledge bases for RAG and tools for external actions. Best for customer support, Q&A, and general assistants.

Workflow App

A visual node-graph workflow for complex, multi-step logic. Chain LLM calls, conditions, loops, and tool invocations. Best for data pipelines, automation, and orchestrated agent behavior.

Channel App

An application connected to external messaging platforms such as Slack, KakaoTalk, or web widgets. Inherits the capabilities of a Chat or Workflow App while adding channel-specific configuration.

​
Application Settings

Every application has the following configurable properties:

​
General

SettingDescription
NameDisplay name shown in the workspace and to end users
DescriptionBrief summary of what the application does
IconVisual identifier for the application
TypeChat, Workflow, or Channel

​
Model Configuration

SettingDefaultDescription
Model—The AI model to use (e.g., gpt-4o, claude-3.5-sonnet)
Temperature0.7Controls randomness. Lower = more deterministic, higher = more creative.
Max Tokens4096Maximum number of tokens in the model’s response
Top P1.0Nucleus sampling parameter
System Prompt—Instructions that define the agent’s persona and behavior

​
Features

FeatureDescription
StreamingEnable real-time token-by-token response delivery via SSE
MemoryConversation memory with configurable window size (number of past messages retained)
ToolsAttach tools (plugins) the agent can invoke during conversation
Knowledge BasesConnect one or more knowledge bases for retrieval-augmented generation (RAG)

​
Application Lifecycle

1

Create

Create a new application from the workspace dashboard. Choose the application type and give it a name.
2

Configure

Select a model, write a system prompt, set parameters, and optionally attach knowledge bases and tools.
3

Test

Use the built-in chat playground to test your application’s behavior. Iterate on prompts and settings.
4

Deploy

Publish the application to make it available via API, embed widget, or messaging channel.
5

Monitor

Track token usage, response times, costs, and conversation quality through the analytics dashboard.

​
Chat App in Detail

A Chat App is the simplest and most common application type. It wraps a single LLM with:
  • System prompt defining the agent’s role and instructions
  • Knowledge bases for RAG — the agent retrieves relevant documents before answering
  • Tools the agent can call (e.g., web search, calculator, custom APIs)
  • Conversation memory to maintain context across turns
User Message → [Memory Recall] → [Knowledge Retrieval] → [LLM + Tools] → Response

​
Workflow App in Detail

A Workflow App uses a visual node graph to define complex execution flows. Each node performs a specific operation, and edges define the data flow between them. Supported node types include:
  • LLM Node — call an AI model with a prompt
  • Condition Node — branch based on a condition
  • Tool Node — invoke an external tool or API
  • Code Node — execute custom Python or JavaScript
  • Knowledge Node — query a knowledge base
  • Loop Node — iterate over a collection
Workflow Apps are ideal when you need deterministic control flow, parallel execution, or multi-agent orchestration that goes beyond a single LLM conversation.

​
Channel App in Detail

A Channel App connects your Chat or Workflow application to an external messaging platform. Nadoo AI handles the channel-specific protocol (webhooks, message formatting, authentication) so your agent logic stays the same. Supported channels include:
  • Web embed widget
  • Slack
  • KakaoTalk
  • Custom webhook endpoints

​
Managing Applications

​
Via the Dashboard

Use the workspace dashboard to create, edit, duplicate, and delete applications. The dashboard provides a visual interface for all settings.

​
Via the API

All application operations are available through the REST API: 
# List applications
GET /api/v1/applications

# Create an application
POST /api/v1/applications

# Update an application
PUT /api/v1/applications/{app_id}

# Delete an application
DELETE /api/v1/applications/{app_id}
See the API Reference for full endpoint documentation.