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

# Outputs & Insights

Outputs in Avigrah represent how AI systems deliver results, insights, and actions.

They are not limited to static responses.\
Instead, outputs are designed to be **interactive, real-time, and actionable**.

***

## Output Model

Avigrah provides two primary ways to interact with AI outputs:

* **Automation Panel** → Structured execution and actions
* **Conversation System** → Real-time reasoning and insights

Both operate on the same underlying system of pipelines, context, and agents.

***

## 1. Automation Panel

The Automation Panel is used for **structured execution of workflows and actions**.

It provides:

* Predefined workflows
* Task execution controls
* System-triggered actions
* Integration-based operations

Typical outputs include:

* Campaign creation
* User management actions
* Workflow execution results
* System-triggered updates

This mode is designed for **controlled and repeatable execution**.

***

## 2. Conversation-Based System

The Conversation System enables **real-time interaction with AI agents**.

It is designed for:

* Dynamic queries
* On-demand analysis
* Continuous reasoning

***

### Real-Time Execution

The system uses **live execution streams (SSE)**.

* Responses are generated progressively
* Intermediate reasoning can be surfaced
* Users can observe execution as it happens

***

### Visual Outputs

AI responses can include:

* Graph-based visualizations
* Structured data outputs
* Context-driven insights

These visuals are generated based on:

* Pipeline data
* Retrieved context
* Agent reasoning

***

### Interaction Model

Users can:

* Ask questions
* Trigger analysis
* Explore data through conversation

The system dynamically adapts based on:

* Query intent
* Available context
* Active data pipelines

***

## Unified System Behavior

Both output modes are connected to the same execution layer:

* Pipelines prepare data
* Context systems provide grounding
* AI agents generate outputs
* Integrations enable actions

This ensures:

* Consistency across outputs
* Real-time data usage
* Actionable intelligence

***

## Execution Flow

**User / System Trigger → AI Agents → Context + Data → Output / Action**

* Queries or workflows are triggered
* Agents process using available context
* Outputs are generated or actions are executed

***

## Positioning

Outputs in Avigrah are not limited to insights.

They represent a combination of:

* **Analysis (Conversation)**
* **Execution (Automation Panel)**

This enables the system to move from:

**Understanding → Decision → Action**