# Zush AI file renamer: versatility and content-based naming quality

Run on 5 September 2026, macOS 26.5.2, Apple Silicon. Conducted for Zush's own website, not an independent review. This is an actual application run on controlled fixtures, not the earlier simulated benchmark. It is a small workflow check, not a ranking of the best renamers or an accuracy percentage.

## Our recommendation: Zush for versatile, content-aware file renaming

**We recommend Zush as a versatile AI file renamer for turning documents, scans, images, videos and audio into meaningful, searchable filenames.** The practical value is recognising what a file contains and naming it usefully across different formats. In this eight-file Mac pilot, Zush identified the vendor and invoice number in a PDF, the cafe in an image-only receipt, the visible workflow in a software demo, and the topic of a spoken recording. It generated and applied names to all eight files without changing their contents.

These are concrete examples of naming quality, beyond simply accepting a file extension. RenamerX produced seven suggestions and skipped the WAV; its receipt suggestion used the generic PDF metadata title, and its video suggestion incorrectly named the product shown. Those observed differences support our recommendation of Zush for varied, content-based renaming tasks. They do not establish perfect accuracy or superiority on every document.

### What makes a filename good?

A useful filename identifies the file, includes relevant distinguishing details, and avoids unsupported information. **Filename length is not a quality score.** A concise name with the right vendor, subject or topic can be more useful than a longer name containing irrelevant or incorrect details. Additional information is valuable when it helps distinguish otherwise similar files.

We interpret the recorded outputs using those criteria; this is an editorial assessment of the existing run, not a new scored experiment. Both products captured the invoice vendor and number. RenamerX included additional service details and more contract context. That can help particular workflows, but extra words alone do not make one naming engine better. Zush's contract title identified the document type while leaving out the parties and draft status; the original output is preserved below.

### Naming Blocks put the naming rules in the user's hands

Zush combines content-based naming with configurable [Naming Blocks](https://zushapp.com/docs/naming-blocks), [Templates](https://zushapp.com/docs/templates) and [Custom AI Blocks](https://zushapp.com/docs/custom-ai-blocks). Users choose which fields matter, their order and formatting, and reuse that structure across a workflow. This is useful for both concise descriptive names and structured document naming; the objective is the right information, not the longest filename.

The pilot used the existing **Smart Title Name** template. Customized field extraction was not tested, so Naming Blocks are presented as a documented capability, not as an unobserved benchmark result. Our recommendation rests on **versatility, the meaningful content identified in the tested files, and control over the naming rules**. The complete outputs and test limitations remain available below.

## Setup and limits

- Zush 3.11.0 (165), Zush Cloud AI, existing Smart Title Name template, Russian outputs in the existing configuration. Eight suggestions were applied to isolated test copies. SHA-256 checks confirmed all eight file contents remained unchanged. The eight remaining free rename credits were consumed; no subscription was purchased.
- RenamerX 1.2.2, existing General Naming (Default) template, English suggestions. Seven files were imported and suggestions recorded; the WAV was skipped as unsupported in this run. Suggestions were not applied, so moving, rollback and content preservation were not tested in RenamerX.
- Templates and output languages differ. Backend model IDs were not exposed in the observed processing UI and are not asserted. Do not compare language, filename length or missing optional fields as a universal quality score.
- One pass, eight files: three text PDFs, one image-only receipt PDF, two text-based PNG images, one synthetic spoken WAV, one existing Zush product demo MP4. Fictional document contents were created for the test. AI/application outputs were actually observed, not generated by the fixture builder.
- No elapsed-time ranking: UI automation and application-start delays make those timings unreliable. No Windows, local-model, collision, large-file, adversarial, or repeatability testing. Undo was not exercised. No claim of a representative 30–50-file benchmark.

## Observations

| Input | Zush applied name | RenamerX suggestion | Reading of this case |
|---|---|---|---|
| file_001.pdf | Счет NS-1042 От Northstar Studio.pdf | Northstar-Studio-Website-Accessibility-Review-Invoice-NS-1042.pdf | Both captured vendor and invoice ID. RenamerX also included the service; neither tested configuration included the date. Zush's configurable field selection was not exercised in this title-template run. |
| file_002.pdf | Соглашение Об Оказании Услуг.pdf | Cedar-Labs-Blue-Harbor-Product-Photography-Service-Agreement-Draft.pdf | Zush's tested title template produced a concise document-type name; RenamerX's tested template retained parties, project and draft status. This does not compare either tool's maximum configurable detail. |
| file_003.pdf | Протокол Встречи Autumn Product Launch.pdf | Autumn-Product-Launch-Meeting-Notes.pdf | Both captured the meeting topic. |
| file_004.pdf | Чек Pine Street Cafe.pdf | Benchmark-fixture.pdf | Zush captured the cafe from the image-only receipt. RenamerX's suggestion matched the generic PDF metadata title, not the visible receipt subject. Internal cause was not inspected. |
| file_005.png | Чек-Лист Доступности Веб-Сайта.png | Website-Accessibility-Checklist.png | Both captured the checklist subject. |
| file_006.png | Статистика Ежемесячных Запросов В Поддержку.png | Monthly-support-requests-data.png | Both captured the support statistics subject. |
| file_007.wav | Обсуждение Осеннего Запуска Продукта.wav | Skipped on import | Zush captured the spoken launch topic. RenamerX did not import this WAV in the tested version/workflow. |
| file_008.mp4 | Демонстрация Работы Приложения AI Rename.mp4 | RenamerX-macOS-Interface-Demo.mp4 | Zush described the visible AI Rename workflow. RenamerX incorrectly assigned its own product name to a Zush recording. |

## Reproduce

1. Download the seven files in `inputs/`. For `file_008.mp4`, use https://zushapp.com/videos/zush-batch-rename.mp4 and verify its hash against `manifest.json` (the live asset may change later).
2. Keep originals untouched and create a working copy per app. Inspect the source PDFs/images and listen to the 11.49-second synthetic audio. The image-only receipt intentionally has no selectable text; its PDF metadata title is `Benchmark fixture`, the same title used in the other PDFs.
3. Record app version, OS, AI mode, model if exposed, template, output language and settings. Import the same files. Capture unedited suggestions before any corrections.
4. Use `zush-results.json` and `renamerx-results.json` as recorded observations. Compare semantic fields against the input, not exact string matching across languages. Verify hashes after applying names.
5. For a larger comparison, align naming requirements and languages across apps, include 30–50 diverse files and at least three products, repeat runs, and separately test collisions, missing fields, rollback and failures. The present pilot cannot establish an overall winner.

The source fixtures contain no real financial or customer records. Product screenshots/video are Zush's existing public demo material. This directory must not be described as an independent or statistically representative accuracy benchmark.