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

# Overview

> What Reeva is, what it does, and the building blocks you'll use.

Reeva is an AI workflow platform for manufacturers. You build **agents** that read engineering documents and structured data from your existing systems — Teamcenter, SAP, file shares — apply AI to extract or classify what matters, and write the results back to the systems your team already uses.

The most common job: keep engineering documents and data **linked and consistent across Teamcenter and SAP**, so your engineers stop rekeying specs and your operators stop chasing the "real" version of a part.

## What you can do with Reeva

* **Link documents across systems.** Match Teamcenter drawings, datasheets, and specs to the corresponding SAP material masters and BOM lines.
* **Extract structured data from documents.** Pull attributes from PDFs, spec sheets, and CAD metadata and write them to the right field in the right system.
* **Validate and route changes.** Detect engineering changes, classify them, and route them through review before they hit downstream systems.
* **Monitor everything.** See what each agent produced, with confidence scores, in one place.

## Core concepts

<CardGroup cols={2}>
  <Card title="Agent" icon="workflow">
    A reusable workflow made of ordered steps. Each agent has a single responsibility — for example, "link new Teamcenter drawings to SAP materials."
  </Card>

  <Card title="Step" icon="square-stack">
    The building block of an agent. Steps come in four types: **trigger**, **action**, **AI**, and **decision**.
  </Card>

  <Card title="Integration" icon="plug">
    A connection to an external system (Teamcenter, SAP, S3, webhook). Agents read from and write to integrations.
  </Card>

  <Card title="Monitor" icon="activity">
    The shared view of every item your agents have produced. Filter by status — published, flagged, linked, queued — and review what needs attention.
  </Card>
</CardGroup>

### Step types

Every agent is made of these four step types, in any order:

| Type         | What it does                                     | Example                                                      |
| ------------ | ------------------------------------------------ | ------------------------------------------------------------ |
| **Trigger**  | Starts the agent when something happens.         | A new drawing is checked into Teamcenter.                    |
| **Action**   | Performs a deterministic task.                   | Fetch the linked spec sheet from a file share.               |
| **AI**       | Uses a model to extract, classify, or summarize. | Extract material grade and tolerances from the spec.         |
| **Decision** | Branches on a condition or confidence threshold. | Auto-publish if confidence ≥ 95%, otherwise flag for review. |

## Where things live in the app

| Section                      | What it's for                                          |
| ---------------------------- | ------------------------------------------------------ |
| **Agents**                   | List, build, and edit agents.                          |
| **Monitor**                  | Review the items your agents have produced.            |
| **Conversations**            | Ask questions about your data and agent runs.          |
| **Settings → Integrations**  | Connect and manage Teamcenter, SAP, and other systems. |
| **Settings → Users & Roles** | Invite teammates and assign permissions.               |
| **Settings → Audit Trail**   | See who changed what, and when.                        |

## Next steps

* Follow the [Quickstart](/get-started/quickstart) to sign in and run your first agent.
* Learn how to [build an agent](/agents/build) step by step.
* Connect [Teamcenter](/integrations/teamcenter) and [SAP](/integrations/sap).