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

# Welcome

> Learn what Phinite is, how it's structured, and how different users can build and manage intelligent agent workflows across environments

## About Phinite

Phinite is an AI automation platform to build, orchestrate, and deploy lifelike assistants and intelligent multi-agent workflows — across channels, tools, and business processes.

Whether you're replacing a call center agent, automating report generation, or deploying assistants that act autonomously, Phinite gives your team the tools to build and run AI systems at scale — safely and visually.

***

## What You Can Do with Phinite

Teams use Phinite to:

* Design agent workflows visually in the Graph Studio
* Build tools and connectors using DevStudio
* Deploy across environments with version control and observability
* Power assistants with knowledge bases, prompt engineering, and tools
* Expose and discover agents via the [Agent Registry](/agent-registry/overview) (A2A protocol)
* Track sessions, model usage, and decisions using detailed logs

***

## Core Concepts

| Concept            | Description                                                                                             |
| ------------------ | ------------------------------------------------------------------------------------------------------- |
| **Assistant**      | A conversational, email-based, or autonomous agent powered by one or more agent graphs                  |
| **Agent Graph**    | A visual decision graph with nodes that orchestrate logic, tools, and conversations                     |
| **Node**           | The building block of a agent graph — can prompt a model, call a tool, capture input, or make decisions |
| **Tool**           | A callable API, function, or system integration used by your assistant                                  |
| **Trigger**        | Defines how an assistant is invoked (e.g., voice call, webhook, calendar event)                         |
| **Environment**    | Isolated stages like Dev, UAT, or Production — with separate assistants, logs, and usage caps           |
| **Session**        | A full execution of an agent graph, recorded in logs with inputs, decisions, outputs, and metrics       |
| **Agent Registry** | Workspace catalog of Agent Cards and hosted A2A endpoints for exposing and composing external agents    |

***

## Who Uses Phinite

Phinite supports a multi-role collaboration model so your entire team can work together:

| Role           | Responsibilities                                                  |
| -------------- | ----------------------------------------------------------------- |
| **Developers** | Build tools, wire integrations, and design agent graphs           |
| **Architects** | Plan agent logic, manage agent graphs, and configure environments |
| **Admins**     | Control access, invite users, and manage billing and environments |
| **Testers**    | Validate assistant behavior and run debug sessions                |
| **Analysts**   | Observe sessions, track model usage, and monitor outcomes         |

***

## What Makes Phinite Different

* Visual-first AI workflow builder, with support for both no-code and code-based design
* Multi-agent orchestration through linked agent graphs and manager agents
* [Agent Registry (A2A)](/agent-registry/overview) for exposing, discovering, and composing external agents
* Deep observability with session-level timelines, decision logs, and token tracking
* Role-based access and environment isolation for enterprise-grade deployments
* Reusable tools, prompts, and assistant templates to reduce repetitive work

***

## Next Steps

* [Quick Start](./quickstart)
* [Workspace Overview](./workspaces/workspace-overview)
* [Inviting Users](./user-management/inviting-users)

***

> Phinite is built for teams that want to go beyond chatbots—enabling AI systems that reason, decide, and act intelligently.