> ## Documentation Index
> Fetch the complete documentation index at: https://docs.subverseai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction to AgentVerse

> SubVerse AI's Agent Orchestration Engine — build and run automated workflows that combine AI agents with business integrations

# Introduction to AgentVerse

AgentVerse is SubVerse AI's Agent Orchestration Engine — a visual, low-code platform for building and running automated workflows that combine AI agents with business integrations.

<img style={{ borderRadius: '0.5rem' }} src="https://mintcdn.com/subverse-611dde60/UZLalX8VX2i72yfQ/images/agentverse/orchestration.png?fit=max&auto=format&n=UZLalX8VX2i72yfQ&q=85&s=2b9ac9999ea09c1f87fcd1174d1c4f91" width="3024" height="1594" data-path="images/agentverse/orchestration.png" />

***

## What is AgentVerse?

AgentVerse lets you design **orchestrations** — directed graphs of nodes that execute in sequence or in parallel. Each node represents a discrete unit of work: calling an API, running an AI model, applying conditional logic, or waiting for a human decision.

Orchestrations are triggered by events (webhooks, schedules, incoming messages) and can route data through AI agents, third-party services, and control-flow logic before delivering a result.

***

## Core Concepts

| Concept               | Description                                                       |
| --------------------- | ----------------------------------------------------------------- |
| **Orchestration**     | The workflow graph you build — a collection of connected nodes    |
| **Node**              | A single step in an orchestration (trigger, action, logic, or AI) |
| **Orchestration Run** | A single execution instance of an orchestration                   |
| **Connection**        | A directed edge between two nodes that carries data               |
| **Credentials**       | Stored authentication details used by integration nodes           |

***

## Node Categories

AgentVerse nodes are organised into four categories:

<CardGroup cols={2}>
  <Card title="Triggers" icon="bolt" href="./node-types">
    Start an orchestration in response to an event — webhook, schedule, Slack, WhatsApp, Shopify, and more
  </Card>

  <Card title="Actions" icon="arrow-right" href="./node-types">
    Interact with external services — send messages, write records, make HTTP calls, upload files
  </Card>

  <Card title="Logic & Flow" icon="code-branch" href="./node-types">
    Control execution — branch with conditions, loop over arrays, pause and wait, require human approval
  </Card>

  <Card title="AI & Agents" icon="robot" href="./node-types">
    Run AI — call any LLM provider or execute full SubVerse AI agent conversations
  </Card>
</CardGroup>

***

## Key Features

* **Visual canvas** — drag-and-drop orchestration builder with real-time previews
* **Any LLM** — connect AI Agent nodes to OpenAI, Anthropic, Google, or any OpenAI-compatible endpoint
* **Dynamic inputs** — reference upstream node outputs anywhere using `{{ nodeName.field }}` expressions
* **Human-in-the-loop** — pause orchestrations for human approval via Slack or email
* **Robust execution engine** — built for reliability with fault tolerance and retry support
* **Multi-channel AI agents** — run SubVerse AI agents over voice, WhatsApp, email, and chat
* **Fine-grained monitoring** — inspect every orchestration run, node output, and error in detail

***

## How Orchestrations Work

```
Trigger Node
    │
    ▼
Action / AI / Logic Node(s)
    │
    ▼
Output (API call, message, database write, etc.)
```

Data flows between nodes as a structured JSON object. Each node receives the output of its upstream node, transforms or acts on it, and passes the result downstream.

Reference any upstream value in a node parameter using double curly braces:

```
{{ webhook.body.message }}
{{ aiAgent.text }}
{{ airtable.fields.Status }}
```

<Tip>
  See [Dynamic Inputs](./dynamic-inputs.md) for a full guide on the `{{ }}` expression syntax and how to link fields using the visual picker.
</Tip>

For AI Agent nodes, **AI Tool** connections (sub-edges) allow the agent to call other nodes as tools during reasoning.

***

## Next Steps

<CardGroup cols={3}>
  <Card title="Quickstart" icon="rocket" href="./quickstart">
    Build your first orchestration in 10 minutes
  </Card>

  <Card title="Node Types" icon="diagram-project" href="./node-types">
    Understand all node categories in detail
  </Card>

  <Card title="Monitoring" icon="chart-line" href="./monitoring">
    Track and debug orchestration runs
  </Card>
</CardGroup>