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

# FineParser Enterprise

> [!WARNING]
>
> FineParser Enterprise is in limited testing with select customers and is not publicly available yet. This documentation is published for participating customers. Features and availability may change before general release.

FineParser Enterprise runs FineParser on Kubernetes as a scalable service. Applications send documents to one HTTP endpoint, and a pool of recognition workers grows and shrinks with the queue. Large PDFs are split into parts, recognized in parallel, and merged back into one result per format.

It ships through AWS Marketplace as a Helm chart, `fineparser-enterprise`, that runs on Amazon EKS. Usage is billed per worker pod-hour through AWS Marketplace.

> [card] [Quickstart](https://docs.abbyy.com/fine-parser/enterprise/getting-started/quickstart.md)
>
> Install the chart and parse your first document.

> [card] [Scaling](https://docs.abbyy.com/fine-parser/enterprise/concepts/scaling.md)
>
> How the worker pool is sized.

> [card] [AWS](https://docs.abbyy.com/fine-parser/enterprise/aws/requirements.md)
>
> Requirements for Amazon EKS.

## Compared with FineParser Basic

| | Basic | Enterprise |
| - | - | - |
| Deployment | One Docker container | Helm chart on Kubernetes |
| Throughput | One document at a time | A worker pool that scales with the queue |
| Large PDFs | Recognized whole | Split and recognized in parallel |
| Recognition settings | Fixed per container | A deployment default, which any request can replace |
| Output formats | DocLang, JSON, plain text | DocLang, JSON, XML, plain text |

## How it works

| Component | Role |
| - | - |
| Ingest | The HTTP API. Accepts documents and serves results. |
| Orchestrator | Holds the job queue and scales the worker pool. Runs as one replica. |
| Worker | Recognizes documents with FineReader Engine. Scales from zero to `scaling.maxWorkers`. |

A client submits documents to ingest and gets a job ID for each. PDFs over 25 pages are split into parts, workers recognize the parts in parallel, and the merged results are written to the output volume. The client polls the job and downloads its results. See [Storage](https://docs.abbyy.com/fine-parser/enterprise/basics/storage.md) for the volumes this uses.